|
{ |
|
"epoch": 1.0, |
|
"eval_MSE/layer0": 220380.4595384912, |
|
"eval_MSE/layer1": 132.78432877894963, |
|
"eval_MSE/layer2": 365.9396393365076, |
|
"eval_MSE/layer3": 415.98040078389045, |
|
"eval_accuracy": 0.4525254617525837, |
|
"eval_dead_code_fraction/layer0": 0.99694, |
|
"eval_dead_code_fraction/layer1": 0.9535, |
|
"eval_dead_code_fraction/layer2": 0.93486, |
|
"eval_dead_code_fraction/layer3": 0.98186, |
|
"eval_first_transition_accuracy": 0.88, |
|
"eval_input_norm/layer0": 333.77172351868165, |
|
"eval_input_norm/layer1": 6.54500140022604, |
|
"eval_input_norm/layer2": 6.137018968109251, |
|
"eval_input_norm/layer3": 7.40972196774554, |
|
"eval_loss": 1.2691402435302734, |
|
"eval_multicode_k": 1, |
|
"eval_output_norm/layer0": 12.936006403074337, |
|
"eval_output_norm/layer1": 13.144865618203756, |
|
"eval_output_norm/layer2": 18.324818944643734, |
|
"eval_output_norm/layer3": 18.466466705456643, |
|
"eval_runtime": 40.0743, |
|
"eval_samples_per_second": 817.682, |
|
"eval_steps_per_second": 1.597, |
|
"eval_transition_accuracy": 0.5633870967741935, |
|
"perplexity": 3.5577924120078235 |
|
} |