laurent.sexy

normal mode unavailable here

root/nn/match-A/out

--- Directories ---

dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu-anim dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp-anim dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc-anim dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu-anim dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp-anim dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc-anim

--- Files ---

dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu_loss_mse.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_cos.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dot.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_dotsim.npy dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp_loss_mse.npy ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_loss-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png ext_raw-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-cos-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dot-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp-new.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-dotsim-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.00390625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.015625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.0625-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=0.25-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero2-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=hetero3-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+cos-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.0001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse+dot-opti=sgd-bwd_lr=0.1-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.0001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=adam-bwd_lr=0.001-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.01-model=loglinexp.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=deep-fc.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=logexpgelu.png loss-mse-dim=[20,30]-dim_sample=1.0-loss=mse-opti=sgd-bwd_lr=0.1-model=loglinexp.png