| { | |
| "best_metric": 0.6364516129032258, | |
| "best_model_checkpoint": "output_toy/checkpoint-15500", | |
| "epoch": 0.775, | |
| "global_step": 15500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.0, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 3.5788, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 13.754842758178711, | |
| "max_norm/layer2": 14.107428550720215, | |
| "max_norm/layer3": 14.075852394104004, | |
| "mean_norm": 8.452343925833702, | |
| "mean_norm/layer0": 8.452010810375214, | |
| "mean_norm/layer1": 8.451448321342468, | |
| "mean_norm/layer2": 8.451122224330902, | |
| "mean_norm/layer3": 8.454794347286224, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.03, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 2.2465, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 13.313672065734863, | |
| "max_norm/layer2": 13.402200698852539, | |
| "max_norm/layer3": 13.972790718078613, | |
| "mean_norm": 8.24767641723156, | |
| "mean_norm/layer0": 8.446629762649536, | |
| "mean_norm/layer1": 8.26882529258728, | |
| "mean_norm/layer2": 8.018843710422516, | |
| "mean_norm/layer3": 8.256406903266907, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.3564852016178642, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.31, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.8386362791061401, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.7699, | |
| "eval_samples_per_second": 803.731, | |
| "eval_steps_per_second": 1.57, | |
| "eval_transition_accuracy": 0.3554838709677419, | |
| "step": 500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.05, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.5981, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 13.26521110534668, | |
| "max_norm/layer2": 13.196029663085938, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.097165614366531, | |
| "mean_norm/layer0": 8.447738409042358, | |
| "mean_norm/layer1": 8.100942492485046, | |
| "mean_norm/layer2": 7.6819451451301575, | |
| "mean_norm/layer3": 8.158036410808563, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4204176054226132, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.58, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.4651859998703003, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.3952, | |
| "eval_samples_per_second": 811.185, | |
| "eval_steps_per_second": 1.584, | |
| "eval_transition_accuracy": 0.5014516129032258, | |
| "step": 1000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.07, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.3928, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 13.196317672729492, | |
| "max_norm/layer2": 12.821307182312012, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.05762867629528, | |
| "mean_norm/layer0": 8.449561476707458, | |
| "mean_norm/layer1": 8.045644223690033, | |
| "mean_norm/layer2": 7.585634410381317, | |
| "mean_norm/layer3": 8.149674594402313, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4378030131182333, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.79, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.354053258895874, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.3967, | |
| "eval_samples_per_second": 811.156, | |
| "eval_steps_per_second": 1.584, | |
| "eval_transition_accuracy": 0.555, | |
| "step": 1500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.1, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.3405, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 12.928963661193848, | |
| "max_norm/layer2": 12.629460334777832, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.056178480386734, | |
| "mean_norm/layer0": 8.451180398464203, | |
| "mean_norm/layer1": 8.046804785728455, | |
| "mean_norm/layer2": 7.5705525279045105, | |
| "mean_norm/layer3": 8.156176209449768, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4426980807086614, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.82, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.3263764381408691, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4538, | |
| "eval_samples_per_second": 810.011, | |
| "eval_steps_per_second": 1.582, | |
| "eval_transition_accuracy": 0.5756451612903226, | |
| "step": 2000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.12, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.3189, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 12.928963661193848, | |
| "max_norm/layer2": 12.610198974609375, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.065410792827606, | |
| "mean_norm/layer0": 8.452252388000488, | |
| "mean_norm/layer1": 8.05913120508194, | |
| "mean_norm/layer2": 7.583977818489075, | |
| "mean_norm/layer3": 8.166281759738922, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4445690245140256, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.86, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.3187371492385864, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.0204, | |
| "eval_samples_per_second": 818.782, | |
| "eval_steps_per_second": 1.599, | |
| "eval_transition_accuracy": 0.5575806451612904, | |
| "step": 2500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.15, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.308, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 12.928963661193848, | |
| "max_norm/layer2": 12.610198974609375, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.076771080493927, | |
| "mean_norm/layer0": 8.453002035617828, | |
| "mean_norm/layer1": 8.070474624633789, | |
| "mean_norm/layer2": 7.606890320777893, | |
| "mean_norm/layer3": 8.176717340946198, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.44684558778297245, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.82, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.3064292669296265, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4873, | |
| "eval_samples_per_second": 809.341, | |
| "eval_steps_per_second": 1.581, | |
| "eval_transition_accuracy": 0.557258064516129, | |
| "step": 3000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.17, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.3009, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 12.928963661193848, | |
| "max_norm/layer2": 12.480420112609863, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.087428167462349, | |
| "mean_norm/layer0": 8.45363199710846, | |
| "mean_norm/layer1": 8.081151723861694, | |
| "mean_norm/layer2": 7.628681242465973, | |
| "mean_norm/layer3": 8.186247706413269, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.44931342658095474, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.87, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2963054180145264, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.5536, | |
| "eval_samples_per_second": 808.018, | |
| "eval_steps_per_second": 1.578, | |
| "eval_transition_accuracy": 0.5762903225806452, | |
| "step": 3500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.2, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2965, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 12.945528030395508, | |
| "max_norm/layer2": 13.130059242248535, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.097855687141418, | |
| "mean_norm/layer0": 8.454240918159485, | |
| "mean_norm/layer1": 8.09092777967453, | |
| "mean_norm/layer2": 7.649804890155792, | |
| "mean_norm/layer3": 8.196449160575867, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4493698961152805, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2922283411026, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.3081, | |
| "eval_samples_per_second": 812.939, | |
| "eval_steps_per_second": 1.588, | |
| "eval_transition_accuracy": 0.567741935483871, | |
| "step": 4000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.23, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2919, | |
| "max_norm": 14.111713409423828, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 13.611448287963867, | |
| "max_norm/layer2": 13.752634048461914, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.10820010304451, | |
| "mean_norm/layer0": 8.454896569252014, | |
| "mean_norm/layer1": 8.099043369293213, | |
| "mean_norm/layer2": 7.671496093273163, | |
| "mean_norm/layer3": 8.20736438035965, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.449889175535187, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.91, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2880299091339111, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.5303, | |
| "eval_samples_per_second": 808.482, | |
| "eval_steps_per_second": 1.579, | |
| "eval_transition_accuracy": 0.5820967741935484, | |
| "step": 4500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.25, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2889, | |
| "max_norm": 14.629088401794434, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 14.212156295776367, | |
| "max_norm/layer2": 14.629088401794434, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.117734983563423, | |
| "mean_norm/layer0": 8.455368101596832, | |
| "mean_norm/layer1": 8.104798018932343, | |
| "mean_norm/layer2": 7.693721830844879, | |
| "mean_norm/layer3": 8.217051982879639, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45008597786970966, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2855565547943115, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.1444, | |
| "eval_samples_per_second": 816.252, | |
| "eval_steps_per_second": 1.594, | |
| "eval_transition_accuracy": 0.56, | |
| "step": 5000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.28, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2855, | |
| "max_norm": 15.594667434692383, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 14.633105278015137, | |
| "max_norm/layer2": 15.594667434692383, | |
| "max_norm/layer3": 13.539477348327637, | |
| "mean_norm": 8.12691231071949, | |
| "mean_norm/layer0": 8.45590054988861, | |
| "mean_norm/layer1": 8.110510230064392, | |
| "mean_norm/layer2": 7.71501624584198, | |
| "mean_norm/layer3": 8.226222217082977, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4503207469549705, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2815940380096436, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4723, | |
| "eval_samples_per_second": 809.64, | |
| "eval_steps_per_second": 1.581, | |
| "eval_transition_accuracy": 0.6016129032258064, | |
| "step": 5500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.3, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2828, | |
| "max_norm": 16.523221969604492, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 15.140296936035156, | |
| "max_norm/layer2": 16.523221969604492, | |
| "max_norm/layer3": 15.213438987731934, | |
| "mean_norm": 8.136623591184616, | |
| "mean_norm/layer0": 8.456406712532043, | |
| "mean_norm/layer1": 8.11625623703003, | |
| "mean_norm/layer2": 7.738001704216003, | |
| "mean_norm/layer3": 8.235829710960388, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45019699457123524, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.87, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2844185829162598, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4554, | |
| "eval_samples_per_second": 809.979, | |
| "eval_steps_per_second": 1.582, | |
| "eval_transition_accuracy": 0.5733870967741935, | |
| "step": 6000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.33, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2805, | |
| "max_norm": 17.3682861328125, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 15.587692260742188, | |
| "max_norm/layer2": 17.3682861328125, | |
| "max_norm/layer3": 16.75887107849121, | |
| "mean_norm": 8.145677655935287, | |
| "mean_norm/layer0": 8.456890940666199, | |
| "mean_norm/layer1": 8.120486974716187, | |
| "mean_norm/layer2": 7.760386824607849, | |
| "mean_norm/layer3": 8.244945883750916, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45155730960875984, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.95, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.277733325958252, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4778, | |
| "eval_samples_per_second": 809.529, | |
| "eval_steps_per_second": 1.581, | |
| "eval_transition_accuracy": 0.6083870967741936, | |
| "step": 6500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.35, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2793, | |
| "max_norm": 18.367448806762695, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 16.05299949645996, | |
| "max_norm/layer2": 18.142847061157227, | |
| "max_norm/layer3": 18.367448806762695, | |
| "mean_norm": 8.15514886379242, | |
| "mean_norm/layer0": 8.45737361907959, | |
| "mean_norm/layer1": 8.127346932888031, | |
| "mean_norm/layer2": 7.781527459621429, | |
| "mean_norm/layer3": 8.254347443580627, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45114255890132876, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.93, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2795634269714355, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.6598, | |
| "eval_samples_per_second": 805.906, | |
| "eval_steps_per_second": 1.574, | |
| "eval_transition_accuracy": 0.5680645161290323, | |
| "step": 7000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.38, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2785, | |
| "max_norm": 20.019760131835938, | |
| "max_norm/layer0": 14.111713409423828, | |
| "max_norm/layer1": 16.42165756225586, | |
| "max_norm/layer2": 19.017709732055664, | |
| "max_norm/layer3": 20.019760131835938, | |
| "mean_norm": 8.164311796426773, | |
| "mean_norm/layer0": 8.457763373851776, | |
| "mean_norm/layer1": 8.13219028711319, | |
| "mean_norm/layer2": 7.803518235683441, | |
| "mean_norm/layer3": 8.263775289058685, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45188651497908466, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.95, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2748253345489502, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4299, | |
| "eval_samples_per_second": 810.489, | |
| "eval_steps_per_second": 1.583, | |
| "eval_transition_accuracy": 0.5919354838709677, | |
| "step": 7500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.4, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2764, | |
| "max_norm": 21.821395874023438, | |
| "max_norm/layer0": 14.158592224121094, | |
| "max_norm/layer1": 16.7973690032959, | |
| "max_norm/layer2": 19.965808868408203, | |
| "max_norm/layer3": 21.821395874023438, | |
| "mean_norm": 8.173083677887917, | |
| "mean_norm/layer0": 8.458155512809753, | |
| "mean_norm/layer1": 8.137401163578033, | |
| "mean_norm/layer2": 7.8241875767707825, | |
| "mean_norm/layer3": 8.272590458393097, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.451844222902313, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.276716709136963, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.3344, | |
| "eval_samples_per_second": 812.407, | |
| "eval_steps_per_second": 1.587, | |
| "eval_transition_accuracy": 0.5759677419354838, | |
| "step": 8000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.42, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2763, | |
| "max_norm": 23.506425857543945, | |
| "max_norm/layer0": 14.451132774353027, | |
| "max_norm/layer1": 17.121503829956055, | |
| "max_norm/layer2": 20.906761169433594, | |
| "max_norm/layer3": 23.506425857543945, | |
| "mean_norm": 8.181086376309395, | |
| "mean_norm/layer0": 8.458472549915314, | |
| "mean_norm/layer1": 8.140588343143463, | |
| "mean_norm/layer2": 7.84406965970993, | |
| "mean_norm/layer3": 8.281214952468872, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45069200410617616, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.94, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2800538539886475, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.8942, | |
| "eval_samples_per_second": 801.287, | |
| "eval_steps_per_second": 1.565, | |
| "eval_transition_accuracy": 0.582741935483871, | |
| "step": 8500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.45, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2755, | |
| "max_norm": 25.07321548461914, | |
| "max_norm/layer0": 14.874269485473633, | |
| "max_norm/layer1": 17.39398956298828, | |
| "max_norm/layer2": 21.67055892944336, | |
| "max_norm/layer3": 25.07321548461914, | |
| "mean_norm": 8.188907638192177, | |
| "mean_norm/layer0": 8.458899140357971, | |
| "mean_norm/layer1": 8.143711388111115, | |
| "mean_norm/layer2": 7.863752484321594, | |
| "mean_norm/layer3": 8.289267539978027, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4516255536417323, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2754778861999512, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.3576, | |
| "eval_samples_per_second": 811.942, | |
| "eval_steps_per_second": 1.586, | |
| "eval_transition_accuracy": 0.5764516129032258, | |
| "step": 9000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.47, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2746, | |
| "max_norm": 26.98938751220703, | |
| "max_norm/layer0": 15.152251243591309, | |
| "max_norm/layer1": 17.66726303100586, | |
| "max_norm/layer2": 22.432802200317383, | |
| "max_norm/layer3": 26.98938751220703, | |
| "mean_norm": 8.197539746761322, | |
| "mean_norm/layer0": 8.4593066573143, | |
| "mean_norm/layer1": 8.148526132106781, | |
| "mean_norm/layer2": 7.884801626205444, | |
| "mean_norm/layer3": 8.297524571418762, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45229261503444884, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2736179828643799, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.4635, | |
| "eval_samples_per_second": 809.817, | |
| "eval_steps_per_second": 1.582, | |
| "eval_transition_accuracy": 0.5864516129032258, | |
| "step": 9500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.5, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2734, | |
| "max_norm": 28.549026489257812, | |
| "max_norm/layer0": 15.429606437683105, | |
| "max_norm/layer1": 17.9191837310791, | |
| "max_norm/layer2": 23.421247482299805, | |
| "max_norm/layer3": 28.549026489257812, | |
| "mean_norm": 8.206039026379585, | |
| "mean_norm/layer0": 8.459603905677795, | |
| "mean_norm/layer1": 8.153472065925598, | |
| "mean_norm/layer2": 7.905435502529144, | |
| "mean_norm/layer3": 8.305644631385803, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4518793061023622, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.91, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2739558219909668, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.3957, | |
| "eval_samples_per_second": 811.175, | |
| "eval_steps_per_second": 1.584, | |
| "eval_transition_accuracy": 0.5779032258064516, | |
| "step": 10000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.53, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2732, | |
| "max_norm": 30.3062801361084, | |
| "max_norm/layer0": 15.69118595123291, | |
| "max_norm/layer1": 18.259578704833984, | |
| "max_norm/layer2": 24.377281188964844, | |
| "max_norm/layer3": 30.3062801361084, | |
| "mean_norm": 8.214424923062325, | |
| "mean_norm/layer0": 8.459942817687988, | |
| "mean_norm/layer1": 8.156991243362427, | |
| "mean_norm/layer2": 7.927173614501953, | |
| "mean_norm/layer3": 8.31359201669693, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45159791961429624, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.89, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2743829488754272, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.5803, | |
| "eval_samples_per_second": 807.485, | |
| "eval_steps_per_second": 1.577, | |
| "eval_transition_accuracy": 0.5879032258064516, | |
| "step": 10500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.55, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2723, | |
| "max_norm": 31.752639770507812, | |
| "max_norm/layer0": 16.03121566772461, | |
| "max_norm/layer1": 18.51296043395996, | |
| "max_norm/layer2": 25.478057861328125, | |
| "max_norm/layer3": 31.752639770507812, | |
| "mean_norm": 8.222758993506432, | |
| "mean_norm/layer0": 8.460269570350647, | |
| "mean_norm/layer1": 8.1603884100914, | |
| "mean_norm/layer2": 7.949049711227417, | |
| "mean_norm/layer3": 8.321328282356262, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45252594234436516, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.89, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.269043207168579, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.1138, | |
| "eval_samples_per_second": 816.875, | |
| "eval_steps_per_second": 1.595, | |
| "eval_transition_accuracy": 0.5811290322580646, | |
| "step": 11000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.57, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2712, | |
| "max_norm": 33.44036102294922, | |
| "max_norm/layer0": 16.178028106689453, | |
| "max_norm/layer1": 18.79450798034668, | |
| "max_norm/layer2": 26.372129440307617, | |
| "max_norm/layer3": 33.44036102294922, | |
| "mean_norm": 8.230609133839607, | |
| "mean_norm/layer0": 8.460545778274536, | |
| "mean_norm/layer1": 8.163512825965881, | |
| "mean_norm/layer2": 7.969985008239746, | |
| "mean_norm/layer3": 8.328392922878265, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45260740265132876, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.93, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2705051898956299, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.2988, | |
| "eval_samples_per_second": 813.126, | |
| "eval_steps_per_second": 1.588, | |
| "eval_transition_accuracy": 0.5779032258064516, | |
| "step": 11500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.6, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2716, | |
| "max_norm": 35.007076263427734, | |
| "max_norm/layer0": 16.497554779052734, | |
| "max_norm/layer1": 18.986597061157227, | |
| "max_norm/layer2": 27.318687438964844, | |
| "max_norm/layer3": 35.007076263427734, | |
| "mean_norm": 8.238363325595856, | |
| "mean_norm/layer0": 8.460874915122986, | |
| "mean_norm/layer1": 8.166603803634644, | |
| "mean_norm/layer2": 7.990897297859192, | |
| "mean_norm/layer3": 8.335077285766602, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4526636318897638, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.89, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2700704336166382, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.2954, | |
| "eval_samples_per_second": 813.195, | |
| "eval_steps_per_second": 1.588, | |
| "eval_transition_accuracy": 0.5759677419354838, | |
| "step": 12000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.62, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2708, | |
| "max_norm": 36.530174255371094, | |
| "max_norm/layer0": 16.774707794189453, | |
| "max_norm/layer1": 19.286115646362305, | |
| "max_norm/layer2": 28.3411808013916, | |
| "max_norm/layer3": 36.530174255371094, | |
| "mean_norm": 8.246006086468697, | |
| "mean_norm/layer0": 8.461170196533203, | |
| "mean_norm/layer1": 8.169004082679749, | |
| "mean_norm/layer2": 8.012158274650574, | |
| "mean_norm/layer3": 8.341691792011261, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45218808632197344, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.95, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2715603113174438, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.6419, | |
| "eval_samples_per_second": 806.261, | |
| "eval_steps_per_second": 1.575, | |
| "eval_transition_accuracy": 0.5485483870967742, | |
| "step": 12500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.65, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2705, | |
| "max_norm": 38.14279556274414, | |
| "max_norm/layer0": 16.91411590576172, | |
| "max_norm/layer1": 19.68568229675293, | |
| "max_norm/layer2": 29.167861938476562, | |
| "max_norm/layer3": 38.14279556274414, | |
| "mean_norm": 8.253673061728477, | |
| "mean_norm/layer0": 8.461421430110931, | |
| "mean_norm/layer1": 8.171639680862427, | |
| "mean_norm/layer2": 8.033495247364044, | |
| "mean_norm/layer3": 8.348135888576508, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4528599536325049, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.93, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2675950527191162, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.2422, | |
| "eval_samples_per_second": 814.27, | |
| "eval_steps_per_second": 1.59, | |
| "eval_transition_accuracy": 0.5733870967741935, | |
| "step": 13000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.68, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2696, | |
| "max_norm": 39.6273307800293, | |
| "max_norm/layer0": 17.14823341369629, | |
| "max_norm/layer1": 19.711994171142578, | |
| "max_norm/layer2": 30.05324363708496, | |
| "max_norm/layer3": 39.6273307800293, | |
| "mean_norm": 8.261498123407364, | |
| "mean_norm/layer0": 8.46165120601654, | |
| "mean_norm/layer1": 8.17492812871933, | |
| "mean_norm/layer2": 8.055188477039337, | |
| "mean_norm/layer3": 8.354224681854248, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4518833911325049, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.91, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2716896533966064, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.327, | |
| "eval_samples_per_second": 812.557, | |
| "eval_steps_per_second": 1.587, | |
| "eval_transition_accuracy": 0.5993548387096774, | |
| "step": 13500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.7, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2687, | |
| "max_norm": 41.1092643737793, | |
| "max_norm/layer0": 17.324541091918945, | |
| "max_norm/layer1": 19.974578857421875, | |
| "max_norm/layer2": 30.87788200378418, | |
| "max_norm/layer3": 41.1092643737793, | |
| "mean_norm": 8.26871033012867, | |
| "mean_norm/layer0": 8.46183955669403, | |
| "mean_norm/layer1": 8.177218735218048, | |
| "mean_norm/layer2": 8.076003432273865, | |
| "mean_norm/layer3": 8.359779596328735, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45239882581815943, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.9, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2687005996704102, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.5627, | |
| "eval_samples_per_second": 807.837, | |
| "eval_steps_per_second": 1.578, | |
| "eval_transition_accuracy": 0.5756451612903226, | |
| "step": 14000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.72, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2685, | |
| "max_norm": 42.766014099121094, | |
| "max_norm/layer0": 17.42141342163086, | |
| "max_norm/layer1": 20.071725845336914, | |
| "max_norm/layer2": 31.913164138793945, | |
| "max_norm/layer3": 42.766014099121094, | |
| "mean_norm": 8.276251748204231, | |
| "mean_norm/layer0": 8.46206510066986, | |
| "mean_norm/layer1": 8.179498374462128, | |
| "mean_norm/layer2": 8.097876787185669, | |
| "mean_norm/layer3": 8.365566730499268, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45213714359313484, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.89, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2709327936172485, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.2398, | |
| "eval_samples_per_second": 814.319, | |
| "eval_steps_per_second": 1.59, | |
| "eval_transition_accuracy": 0.612741935483871, | |
| "step": 14500 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.75, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2685, | |
| "max_norm": 44.61077117919922, | |
| "max_norm/layer0": 17.665727615356445, | |
| "max_norm/layer1": 20.15743637084961, | |
| "max_norm/layer2": 32.966209411621094, | |
| "max_norm/layer3": 44.61077117919922, | |
| "mean_norm": 8.283790707588196, | |
| "mean_norm/layer0": 8.462257981300354, | |
| "mean_norm/layer1": 8.181303024291992, | |
| "mean_norm/layer2": 8.120182931423187, | |
| "mean_norm/layer3": 8.37141889333725, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.4519187146284449, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.91, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2706036567687988, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 40.0952, | |
| "eval_samples_per_second": 817.256, | |
| "eval_steps_per_second": 1.596, | |
| "eval_transition_accuracy": 0.587258064516129, | |
| "step": 15000 | |
| }, | |
| { | |
| "MSE": 0.0, | |
| "MSE/layer0": 0.0, | |
| "MSE/layer1": 0.0, | |
| "MSE/layer2": 0.0, | |
| "MSE/layer3": 0.0, | |
| "dead_code_fraction": 1.0, | |
| "dead_code_fraction/layer0": 1.0, | |
| "dead_code_fraction/layer1": 1.0, | |
| "dead_code_fraction/layer2": 1.0, | |
| "dead_code_fraction/layer3": 1.0, | |
| "epoch": 0.78, | |
| "input_norm": 0.0, | |
| "input_norm/layer0": 0.0, | |
| "input_norm/layer1": 0.0, | |
| "input_norm/layer2": 0.0, | |
| "input_norm/layer3": 0.0, | |
| "learning_rate": 0.001, | |
| "loss": 1.2675, | |
| "max_norm": 46.33829879760742, | |
| "max_norm/layer0": 17.856664657592773, | |
| "max_norm/layer1": 20.084186553955078, | |
| "max_norm/layer2": 33.940242767333984, | |
| "max_norm/layer3": 46.33829879760742, | |
| "mean_norm": 8.291451185941696, | |
| "mean_norm/layer0": 8.462452054023743, | |
| "mean_norm/layer1": 8.18280303478241, | |
| "mean_norm/layer2": 8.143204748630524, | |
| "mean_norm/layer3": 8.377344906330109, | |
| "multicode_k": 1, | |
| "output_norm": 0.0, | |
| "output_norm/layer0": 0.0, | |
| "output_norm/layer1": 0.0, | |
| "output_norm/layer2": 0.0, | |
| "output_norm/layer3": 0.0, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "eval_MSE/layer0": 0.0, | |
| "eval_MSE/layer1": 0.0, | |
| "eval_MSE/layer2": 0.0, | |
| "eval_MSE/layer3": 0.0, | |
| "eval_accuracy": 0.45268790177472934, | |
| "eval_dead_code_fraction/layer0": 1.0, | |
| "eval_dead_code_fraction/layer1": 1.0, | |
| "eval_dead_code_fraction/layer2": 1.0, | |
| "eval_dead_code_fraction/layer3": 1.0, | |
| "eval_first_transition_accuracy": 0.96, | |
| "eval_input_norm/layer0": 0.0, | |
| "eval_input_norm/layer1": 0.0, | |
| "eval_input_norm/layer2": 0.0, | |
| "eval_input_norm/layer3": 0.0, | |
| "eval_loss": 1.2690919637680054, | |
| "eval_multicode_k": 1, | |
| "eval_output_norm/layer0": 0.0, | |
| "eval_output_norm/layer1": 0.0, | |
| "eval_output_norm/layer2": 0.0, | |
| "eval_output_norm/layer3": 0.0, | |
| "eval_runtime": 39.8501, | |
| "eval_samples_per_second": 822.281, | |
| "eval_steps_per_second": 1.606, | |
| "eval_transition_accuracy": 0.6364516129032258, | |
| "step": 15500 | |
| } | |
| ], | |
| "max_steps": 20000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 9712749772800000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |