all_results.json

{
    "MSE": 0.0,
    "MSE/layer0": 0.0,
    "dead_code_fraction": 1.0,
    "dead_code_fraction/layer0": 1.0,
    "epoch": 2.04,
    "eval_MSE/layer0": 501.8128262733759,
    "eval_accuracy": 0.5193506309245984,
    "eval_dead_code_fraction/layer0": 0.16395,
    "eval_input_norm/layer0": 31.998895487949337,
    "eval_loss": 2.0353407859802246,
    "eval_multicode_k": 1,
    "eval_output_norm/layer0": 22.80092038433711,
    "eval_runtime": 159.0824,
    "eval_samples_per_second": 29.06,
    "eval_steps_per_second": 1.817,
    "input_norm": 0.0,
    "input_norm/layer0": 0.0,
    "loss": 2.205516522693634,
    "max_norm": 220.5869903564453,
    "max_norm/layer0": 220.5869903564453,
    "mean_norm": 69.14492416381836,
    "mean_norm/layer0": 69.14492416381836,
    "multicode_k": 1,
    "output_norm": 0.0,
    "output_norm/layer0": 0.0,
    "perplexity": 7.6548603501181,
    "runtime": 15654.0479,
    "samples_per_second": 61.326,
    "steps_per_second": 0.639
}