diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4140 @@ +{ + "best_metric": 1.89570152759552, + "best_model_checkpoint": "/tmp/wandb/run-20240207_044253-56k3p8kp/files/train_output/checkpoint-10000", + "epoch": 1.044022968505307, + "eval_steps": 500, + "global_step": 10000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "MSE": 872.5187733968098, + "MSE/layer0": 872.5187733968098, + "dead_code_fraction": 0.0276, + "dead_code_fraction/layer0": 0.0276, + "epoch": 0.0, + "input_norm": 31.997111479441326, + "input_norm/layer0": 31.997111479441326, + "learning_rate": 1e-06, + "loss": 9.0051, + "max_norm": 34.71393966674805, + "max_norm/layer0": 34.71393966674805, + "mean_norm": 31.98521327972412, + "mean_norm/layer0": 31.98521327972412, + "multicode_k": 1, + "output_norm": 4.134780248006185, + "output_norm/layer0": 4.134780248006185, + "step": 1 + }, + { + "MSE": 871.4381560241286, + "MSE/layer0": 871.4381560241286, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.01, + "input_norm": 31.99644809839677, + "input_norm/layer0": 31.99644809839677, + "learning_rate": 5e-05, + "loss": 7.0703, + "max_norm": 34.72187423706055, + "max_norm/layer0": 34.72187423706055, + "mean_norm": 31.991936683654785, + "mean_norm/layer0": 31.991936683654785, + "multicode_k": 1, + "output_norm": 4.145846879401173, + "output_norm/layer0": 4.145846879401173, + "step": 50 + }, + { + "MSE": 868.4475470987957, + "MSE/layer0": 868.4475470987957, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.01, + "input_norm": 31.995786774953213, + "input_norm/layer0": 31.995786774953213, + "learning_rate": 0.0001, + "loss": 4.1515, + "max_norm": 34.748802185058594, + "max_norm/layer0": 34.748802185058594, + "mean_norm": 32.0172176361084, + "mean_norm/layer0": 32.0172176361084, + "multicode_k": 1, + "output_norm": 4.178660261631009, + "output_norm/layer0": 4.178660261631009, + "step": 100 + }, + { + "MSE": 864.7878089396156, + "MSE/layer0": 864.7878089396156, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.02, + "input_norm": 31.995868380864444, + "input_norm/layer0": 31.995868380864444, + "learning_rate": 0.00015, + "loss": 3.596, + "max_norm": 34.7879753112793, + "max_norm/layer0": 34.7879753112793, + "mean_norm": 32.057809829711914, + "mean_norm/layer0": 32.057809829711914, + "multicode_k": 1, + "output_norm": 4.227458424568177, + "output_norm/layer0": 4.227458424568177, + "step": 150 + }, + { + "MSE": 862.2720657348631, + "MSE/layer0": 862.2720657348631, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.02, + "input_norm": 31.996261011759444, + "input_norm/layer0": 31.996261011759444, + "learning_rate": 0.0002, + "loss": 3.3864, + "max_norm": 34.834476470947266, + "max_norm/layer0": 34.834476470947266, + "mean_norm": 32.09993934631348, + "mean_norm/layer0": 32.09993934631348, + "multicode_k": 1, + "output_norm": 4.271083230972291, + "output_norm/layer0": 4.271083230972291, + "step": 200 + }, + { + "MSE": 860.8860168457031, + "MSE/layer0": 860.8860168457031, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.03, + "input_norm": 31.99663330396016, + "input_norm/layer0": 31.99663330396016, + "learning_rate": 0.00025, + "loss": 3.1841, + "max_norm": 34.880577087402344, + "max_norm/layer0": 34.880577087402344, + "mean_norm": 32.15042304992676, + "mean_norm/layer0": 32.15042304992676, + "multicode_k": 1, + "output_norm": 4.302526236375174, + "output_norm/layer0": 4.302526236375174, + "step": 250 + }, + { + "MSE": 859.4145241292313, + "MSE/layer0": 859.4145241292313, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.03, + "input_norm": 31.99707999547323, + "input_norm/layer0": 31.99707999547323, + "learning_rate": 0.0003, + "loss": 2.9941, + "max_norm": 34.94011688232422, + "max_norm/layer0": 34.94011688232422, + "mean_norm": 32.21405220031738, + "mean_norm/layer0": 32.21405220031738, + "multicode_k": 1, + "output_norm": 4.340623443921407, + "output_norm/layer0": 4.340623443921407, + "step": 300 + }, + { + "MSE": 857.4514228312173, + "MSE/layer0": 857.4514228312173, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.04, + "input_norm": 31.997263495127353, + "input_norm/layer0": 31.997263495127353, + "learning_rate": 0.00035, + "loss": 2.8154, + "max_norm": 35.02033996582031, + "max_norm/layer0": 35.02033996582031, + "mean_norm": 32.2895393371582, + "mean_norm/layer0": 32.2895393371582, + "multicode_k": 1, + "output_norm": 4.388785634040833, + "output_norm/layer0": 4.388785634040833, + "step": 350 + }, + { + "MSE": 855.6023776245115, + "MSE/layer0": 855.6023776245115, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.04, + "input_norm": 31.997391548156735, + "input_norm/layer0": 31.997391548156735, + "learning_rate": 0.0004, + "loss": 2.6472, + "max_norm": 35.093902587890625, + "max_norm/layer0": 35.093902587890625, + "mean_norm": 32.36477088928223, + "mean_norm/layer0": 32.36477088928223, + "multicode_k": 1, + "output_norm": 4.438224600950877, + "output_norm/layer0": 4.438224600950877, + "step": 400 + }, + { + "MSE": 852.2393107096357, + "MSE/layer0": 852.2393107096357, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.05, + "input_norm": 31.997483587265002, + "input_norm/layer0": 31.997483587265002, + "learning_rate": 0.00045000000000000004, + "loss": 2.5584, + "max_norm": 35.304176330566406, + "max_norm/layer0": 35.304176330566406, + "mean_norm": 32.54551696777344, + "mean_norm/layer0": 32.54551696777344, + "multicode_k": 1, + "output_norm": 4.531697844664256, + "output_norm/layer0": 4.531697844664256, + "step": 450 + }, + { + "MSE": 845.160081481933, + "MSE/layer0": 845.160081481933, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.05, + "input_norm": 31.997576513290404, + "input_norm/layer0": 31.997576513290404, + "learning_rate": 0.0005, + "loss": 2.5072, + "max_norm": 35.78097915649414, + "max_norm/layer0": 35.78097915649414, + "mean_norm": 32.836992263793945, + "mean_norm/layer0": 32.836992263793945, + "multicode_k": 1, + "output_norm": 4.75731077671051, + "output_norm/layer0": 4.75731077671051, + "step": 500 + }, + { + "epoch": 0.05, + "eval_MSE/layer0": 841.1602262364518, + "eval_accuracy": 0.4578774282778804, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.99765928777141, + "eval_loss": 2.476405382156372, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 4.911408371361153, + "eval_runtime": 73.5499, + "eval_samples_per_second": 62.855, + "eval_steps_per_second": 7.859, + "step": 500 + }, + { + "MSE": 837.1320628865564, + "MSE/layer0": 837.1320628865564, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.06, + "input_norm": 31.997703491846714, + "input_norm/layer0": 31.997703491846714, + "learning_rate": 0.0004973684210526315, + "loss": 2.446, + "max_norm": 36.301849365234375, + "max_norm/layer0": 36.301849365234375, + "mean_norm": 33.16576957702637, + "mean_norm/layer0": 33.16576957702637, + "multicode_k": 1, + "output_norm": 5.083427506287892, + "output_norm/layer0": 5.083427506287892, + "step": 550 + }, + { + "MSE": 829.8174697875975, + "MSE/layer0": 829.8174697875975, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.06, + "input_norm": 31.997781289418548, + "input_norm/layer0": 31.997781289418548, + "learning_rate": 0.0004947368421052632, + "loss": 2.4026, + "max_norm": 36.790077209472656, + "max_norm/layer0": 36.790077209472656, + "mean_norm": 33.519426345825195, + "mean_norm/layer0": 33.519426345825195, + "multicode_k": 1, + "output_norm": 5.438902084827422, + "output_norm/layer0": 5.438902084827422, + "step": 600 + }, + { + "MSE": 823.647299601237, + "MSE/layer0": 823.647299601237, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.07, + "input_norm": 31.997854344050104, + "input_norm/layer0": 31.997854344050104, + "learning_rate": 0.0004921052631578947, + "loss": 2.3506, + "max_norm": 37.23988723754883, + "max_norm/layer0": 37.23988723754883, + "mean_norm": 33.882219314575195, + "mean_norm/layer0": 33.882219314575195, + "multicode_k": 1, + "output_norm": 5.780141766071318, + "output_norm/layer0": 5.780141766071318, + "step": 650 + }, + { + "MSE": 818.3900874837236, + "MSE/layer0": 818.3900874837236, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.07, + "input_norm": 31.99789853731792, + "input_norm/layer0": 31.99789853731792, + "learning_rate": 0.0004894736842105264, + "loss": 2.3252, + "max_norm": 37.74921417236328, + "max_norm/layer0": 37.74921417236328, + "mean_norm": 34.241193771362305, + "mean_norm/layer0": 34.241193771362305, + "multicode_k": 1, + "output_norm": 6.09345253547033, + "output_norm/layer0": 6.09345253547033, + "step": 700 + }, + { + "MSE": 813.5141651407878, + "MSE/layer0": 813.5141651407878, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.08, + "input_norm": 31.99791718482971, + "input_norm/layer0": 31.99791718482971, + "learning_rate": 0.0004868421052631579, + "loss": 2.2972, + "max_norm": 38.29411315917969, + "max_norm/layer0": 38.29411315917969, + "mean_norm": 34.602651596069336, + "mean_norm/layer0": 34.602651596069336, + "multicode_k": 1, + "output_norm": 6.373116828600564, + "output_norm/layer0": 6.373116828600564, + "step": 750 + }, + { + "MSE": 808.9583784993486, + "MSE/layer0": 808.9583784993486, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.08, + "input_norm": 31.997929916381842, + "input_norm/layer0": 31.997929916381842, + "learning_rate": 0.0004842105263157895, + "loss": 2.2848, + "max_norm": 38.83885955810547, + "max_norm/layer0": 38.83885955810547, + "mean_norm": 34.96581268310547, + "mean_norm/layer0": 34.96581268310547, + "multicode_k": 1, + "output_norm": 6.6348445963859515, + "output_norm/layer0": 6.6348445963859515, + "step": 800 + }, + { + "MSE": 805.0894353230792, + "MSE/layer0": 805.0894353230792, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.09, + "input_norm": 31.99793632825216, + "input_norm/layer0": 31.99793632825216, + "learning_rate": 0.00048157894736842105, + "loss": 2.2718, + "max_norm": 39.34720993041992, + "max_norm/layer0": 39.34720993041992, + "mean_norm": 35.32806396484375, + "mean_norm/layer0": 35.32806396484375, + "multicode_k": 1, + "output_norm": 6.866891795794173, + "output_norm/layer0": 6.866891795794173, + "step": 850 + }, + { + "MSE": 801.1131992594401, + "MSE/layer0": 801.1131992594401, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.09, + "input_norm": 31.997941767374677, + "input_norm/layer0": 31.997941767374677, + "learning_rate": 0.00047894736842105264, + "loss": 2.2552, + "max_norm": 39.885169982910156, + "max_norm/layer0": 39.885169982910156, + "mean_norm": 35.689327239990234, + "mean_norm/layer0": 35.689327239990234, + "multicode_k": 1, + "output_norm": 7.08060004631678, + "output_norm/layer0": 7.08060004631678, + "step": 900 + }, + { + "MSE": 797.5655348714191, + "MSE/layer0": 797.5655348714191, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.1, + "input_norm": 31.997945496241247, + "input_norm/layer0": 31.997945496241247, + "learning_rate": 0.0004763157894736842, + "loss": 2.2364, + "max_norm": 40.398529052734375, + "max_norm/layer0": 40.398529052734375, + "mean_norm": 36.051015853881836, + "mean_norm/layer0": 36.051015853881836, + "multicode_k": 1, + "output_norm": 7.280441036224362, + "output_norm/layer0": 7.280441036224362, + "step": 950 + }, + { + "MSE": 794.0057167561844, + "MSE/layer0": 794.0057167561844, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.1, + "input_norm": 31.997958205540975, + "input_norm/layer0": 31.997958205540975, + "learning_rate": 0.00047368421052631577, + "loss": 2.2285, + "max_norm": 40.882999420166016, + "max_norm/layer0": 40.882999420166016, + "mean_norm": 36.412479400634766, + "mean_norm/layer0": 36.412479400634766, + "multicode_k": 1, + "output_norm": 7.463625483512881, + "output_norm/layer0": 7.463625483512881, + "step": 1000 + }, + { + "epoch": 0.1, + "eval_MSE/layer0": 792.3022871601257, + "eval_accuracy": 0.49262569806414397, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997970815399036, + "eval_loss": 2.2265193462371826, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 7.55243569582849, + "eval_runtime": 73.9102, + "eval_samples_per_second": 62.549, + "eval_steps_per_second": 7.82, + "step": 1000 + }, + { + "MSE": 790.4031213378905, + "MSE/layer0": 790.4031213378905, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.11, + "input_norm": 31.997961893081662, + "input_norm/layer0": 31.997961893081662, + "learning_rate": 0.0004710526315789474, + "loss": 2.2276, + "max_norm": 41.373714447021484, + "max_norm/layer0": 41.373714447021484, + "mean_norm": 36.77394676208496, + "mean_norm/layer0": 36.77394676208496, + "multicode_k": 1, + "output_norm": 7.636834317048386, + "output_norm/layer0": 7.636834317048386, + "step": 1050 + }, + { + "MSE": 786.9933625284832, + "MSE/layer0": 786.9933625284832, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.11, + "input_norm": 31.99796496391297, + "input_norm/layer0": 31.99796496391297, + "learning_rate": 0.00046842105263157895, + "loss": 2.2167, + "max_norm": 41.845481872558594, + "max_norm/layer0": 41.845481872558594, + "mean_norm": 37.13482093811035, + "mean_norm/layer0": 37.13482093811035, + "multicode_k": 1, + "output_norm": 7.803330462773646, + "output_norm/layer0": 7.803330462773646, + "step": 1100 + }, + { + "MSE": 783.8570914713541, + "MSE/layer0": 783.8570914713541, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.12, + "input_norm": 31.997962007522577, + "input_norm/layer0": 31.997962007522577, + "learning_rate": 0.00046578947368421054, + "loss": 2.2049, + "max_norm": 42.328094482421875, + "max_norm/layer0": 42.328094482421875, + "mean_norm": 37.49737358093262, + "mean_norm/layer0": 37.49737358093262, + "multicode_k": 1, + "output_norm": 7.957673575878145, + "output_norm/layer0": 7.957673575878145, + "step": 1150 + }, + { + "MSE": 780.325506286621, + "MSE/layer0": 780.325506286621, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.13, + "input_norm": 31.997955818176273, + "input_norm/layer0": 31.997955818176273, + "learning_rate": 0.00046315789473684214, + "loss": 2.2048, + "max_norm": 42.827125549316406, + "max_norm/layer0": 42.827125549316406, + "mean_norm": 37.85981369018555, + "mean_norm/layer0": 37.85981369018555, + "multicode_k": 1, + "output_norm": 8.110501464207967, + "output_norm/layer0": 8.110501464207967, + "step": 1200 + }, + { + "MSE": 777.4963677978517, + "MSE/layer0": 777.4963677978517, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.13, + "input_norm": 31.997957773208608, + "input_norm/layer0": 31.997957773208608, + "learning_rate": 0.0004605263157894737, + "loss": 2.1813, + "max_norm": 43.32162094116211, + "max_norm/layer0": 43.32162094116211, + "mean_norm": 38.223052978515625, + "mean_norm/layer0": 38.223052978515625, + "multicode_k": 1, + "output_norm": 8.244436805248263, + "output_norm/layer0": 8.244436805248263, + "step": 1250 + }, + { + "MSE": 774.260437520345, + "MSE/layer0": 774.260437520345, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.14, + "input_norm": 31.99796381632487, + "input_norm/layer0": 31.99796381632487, + "learning_rate": 0.00045789473684210527, + "loss": 2.1836, + "max_norm": 43.81217575073242, + "max_norm/layer0": 43.81217575073242, + "mean_norm": 38.58406066894531, + "mean_norm/layer0": 38.58406066894531, + "multicode_k": 1, + "output_norm": 8.38570425987244, + "output_norm/layer0": 8.38570425987244, + "step": 1300 + }, + { + "MSE": 771.4710861206056, + "MSE/layer0": 771.4710861206056, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.14, + "input_norm": 31.997958866755184, + "input_norm/layer0": 31.997958866755184, + "learning_rate": 0.00045526315789473686, + "loss": 2.1749, + "max_norm": 44.29291915893555, + "max_norm/layer0": 44.29291915893555, + "mean_norm": 38.94841957092285, + "mean_norm/layer0": 38.94841957092285, + "multicode_k": 1, + "output_norm": 8.50825534900029, + "output_norm/layer0": 8.50825534900029, + "step": 1350 + }, + { + "MSE": 768.6556185913084, + "MSE/layer0": 768.6556185913084, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.15, + "input_norm": 31.99795674959818, + "input_norm/layer0": 31.99795674959818, + "learning_rate": 0.00045263157894736845, + "loss": 2.1767, + "max_norm": 44.80799865722656, + "max_norm/layer0": 44.80799865722656, + "mean_norm": 39.31004524230957, + "mean_norm/layer0": 39.31004524230957, + "multicode_k": 1, + "output_norm": 8.633222222328184, + "output_norm/layer0": 8.633222222328184, + "step": 1400 + }, + { + "MSE": 765.9088921101885, + "MSE/layer0": 765.9088921101885, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.15, + "input_norm": 31.99795736630759, + "input_norm/layer0": 31.99795736630759, + "learning_rate": 0.00045000000000000004, + "loss": 2.1614, + "max_norm": 45.24712371826172, + "max_norm/layer0": 45.24712371826172, + "mean_norm": 39.66674041748047, + "mean_norm/layer0": 39.66674041748047, + "multicode_k": 1, + "output_norm": 8.743508942921961, + "output_norm/layer0": 8.743508942921961, + "step": 1450 + }, + { + "MSE": 763.2800780232742, + "MSE/layer0": 763.2800780232742, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.16, + "input_norm": 31.997952944437664, + "input_norm/layer0": 31.997952944437664, + "learning_rate": 0.0004473684210526316, + "loss": 2.1472, + "max_norm": 45.6886100769043, + "max_norm/layer0": 45.6886100769043, + "mean_norm": 40.02728462219238, + "mean_norm/layer0": 40.02728462219238, + "multicode_k": 1, + "output_norm": 8.859908480644224, + "output_norm/layer0": 8.859908480644224, + "step": 1500 + }, + { + "epoch": 0.16, + "eval_MSE/layer0": 761.8682555426203, + "eval_accuracy": 0.502513147213907, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.99796608230291, + "eval_loss": 2.1583588123321533, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 8.92388377993132, + "eval_runtime": 73.3386, + "eval_samples_per_second": 63.036, + "eval_steps_per_second": 7.881, + "step": 1500 + }, + { + "MSE": 760.1600253295896, + "MSE/layer0": 760.1600253295896, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.16, + "input_norm": 31.997961203257255, + "input_norm/layer0": 31.997961203257255, + "learning_rate": 0.00044473684210526317, + "loss": 2.1601, + "max_norm": 46.172386169433594, + "max_norm/layer0": 46.172386169433594, + "mean_norm": 40.38890838623047, + "mean_norm/layer0": 40.38890838623047, + "multicode_k": 1, + "output_norm": 8.976485926310215, + "output_norm/layer0": 8.976485926310215, + "step": 1550 + }, + { + "MSE": 757.7968755086266, + "MSE/layer0": 757.7968755086266, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.17, + "input_norm": 31.99795768419901, + "input_norm/layer0": 31.99795768419901, + "learning_rate": 0.0004421052631578947, + "loss": 2.1503, + "max_norm": 46.59892272949219, + "max_norm/layer0": 46.59892272949219, + "mean_norm": 40.74970626831055, + "mean_norm/layer0": 40.74970626831055, + "multicode_k": 1, + "output_norm": 9.079196619192757, + "output_norm/layer0": 9.079196619192757, + "step": 1600 + }, + { + "MSE": 755.1489293416341, + "MSE/layer0": 755.1489293416341, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.17, + "input_norm": 31.997956597010287, + "input_norm/layer0": 31.997956597010287, + "learning_rate": 0.0004394736842105263, + "loss": 2.1474, + "max_norm": 47.01366424560547, + "max_norm/layer0": 47.01366424560547, + "mean_norm": 41.107492446899414, + "mean_norm/layer0": 41.107492446899414, + "multicode_k": 1, + "output_norm": 9.18502354939779, + "output_norm/layer0": 9.18502354939779, + "step": 1650 + }, + { + "MSE": 752.7132907104492, + "MSE/layer0": 752.7132907104492, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.18, + "input_norm": 31.997961333592727, + "input_norm/layer0": 31.997961333592727, + "learning_rate": 0.00043684210526315795, + "loss": 2.1451, + "max_norm": 47.46398162841797, + "max_norm/layer0": 47.46398162841797, + "mean_norm": 41.466739654541016, + "mean_norm/layer0": 41.466739654541016, + "multicode_k": 1, + "output_norm": 9.288365476131446, + "output_norm/layer0": 9.288365476131446, + "step": 1700 + }, + { + "MSE": 750.1894300333656, + "MSE/layer0": 750.1894300333656, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.18, + "input_norm": 31.99795596122742, + "input_norm/layer0": 31.99795596122742, + "learning_rate": 0.0004342105263157895, + "loss": 2.1298, + "max_norm": 47.89784240722656, + "max_norm/layer0": 47.89784240722656, + "mean_norm": 41.825233459472656, + "mean_norm/layer0": 41.825233459472656, + "multicode_k": 1, + "output_norm": 9.383608838717148, + "output_norm/layer0": 9.383608838717148, + "step": 1750 + }, + { + "MSE": 747.6542997233073, + "MSE/layer0": 747.6542997233073, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.19, + "input_norm": 31.997955916722606, + "input_norm/layer0": 31.997955916722606, + "learning_rate": 0.0004315789473684211, + "loss": 2.1238, + "max_norm": 48.32524871826172, + "max_norm/layer0": 48.32524871826172, + "mean_norm": 42.18182373046875, + "mean_norm/layer0": 42.18182373046875, + "multicode_k": 1, + "output_norm": 9.481378455162048, + "output_norm/layer0": 9.481378455162048, + "step": 1800 + }, + { + "MSE": 745.4623332722983, + "MSE/layer0": 745.4623332722983, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.19, + "input_norm": 31.99795308430989, + "input_norm/layer0": 31.99795308430989, + "learning_rate": 0.0004289473684210526, + "loss": 2.1193, + "max_norm": 48.75049591064453, + "max_norm/layer0": 48.75049591064453, + "mean_norm": 42.53817176818848, + "mean_norm/layer0": 42.53817176818848, + "multicode_k": 1, + "output_norm": 9.570223178863522, + "output_norm/layer0": 9.570223178863522, + "step": 1850 + }, + { + "MSE": 743.2356170654296, + "MSE/layer0": 743.2356170654296, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.2, + "input_norm": 31.997956037521366, + "input_norm/layer0": 31.997956037521366, + "learning_rate": 0.0004263157894736842, + "loss": 2.114, + "max_norm": 49.169532775878906, + "max_norm/layer0": 49.169532775878906, + "mean_norm": 42.89301300048828, + "mean_norm/layer0": 42.89301300048828, + "multicode_k": 1, + "output_norm": 9.656177865664167, + "output_norm/layer0": 9.656177865664167, + "step": 1900 + }, + { + "MSE": 740.6696187337238, + "MSE/layer0": 740.6696187337238, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.2, + "input_norm": 31.997947629292796, + "input_norm/layer0": 31.997947629292796, + "learning_rate": 0.0004236842105263158, + "loss": 2.1208, + "max_norm": 49.5915641784668, + "max_norm/layer0": 49.5915641784668, + "mean_norm": 43.247257232666016, + "mean_norm/layer0": 43.247257232666016, + "multicode_k": 1, + "output_norm": 9.750187404950456, + "output_norm/layer0": 9.750187404950456, + "step": 1950 + }, + { + "MSE": 738.2711766560866, + "MSE/layer0": 738.2711766560866, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.21, + "input_norm": 31.99795049031576, + "input_norm/layer0": 31.99795049031576, + "learning_rate": 0.00042105263157894734, + "loss": 2.1144, + "max_norm": 50.01121520996094, + "max_norm/layer0": 50.01121520996094, + "mean_norm": 43.60071563720703, + "mean_norm/layer0": 43.60071563720703, + "multicode_k": 1, + "output_norm": 9.839046444892887, + "output_norm/layer0": 9.839046444892887, + "step": 2000 + }, + { + "epoch": 0.21, + "eval_MSE/layer0": 737.1842960305685, + "eval_accuracy": 0.5089533842961654, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997949216728358, + "eval_loss": 2.112781524658203, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 9.899169789850005, + "eval_runtime": 73.2721, + "eval_samples_per_second": 63.094, + "eval_steps_per_second": 7.888, + "step": 2000 + }, + { + "MSE": 736.3252647908528, + "MSE/layer0": 736.3252647908528, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.21, + "input_norm": 31.997952928543082, + "input_norm/layer0": 31.997952928543082, + "learning_rate": 0.000418421052631579, + "loss": 2.1054, + "max_norm": 50.480525970458984, + "max_norm/layer0": 50.480525970458984, + "mean_norm": 43.9530086517334, + "mean_norm/layer0": 43.9530086517334, + "multicode_k": 1, + "output_norm": 9.923008087476088, + "output_norm/layer0": 9.923008087476088, + "step": 2050 + }, + { + "MSE": 734.2413449096682, + "MSE/layer0": 734.2413449096682, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.22, + "input_norm": 31.99795233090719, + "input_norm/layer0": 31.99795233090719, + "learning_rate": 0.0004157894736842106, + "loss": 2.114, + "max_norm": 50.909828186035156, + "max_norm/layer0": 50.909828186035156, + "mean_norm": 44.302608489990234, + "mean_norm/layer0": 44.302608489990234, + "multicode_k": 1, + "output_norm": 9.99465080579122, + "output_norm/layer0": 9.99465080579122, + "step": 2100 + }, + { + "MSE": 732.1211085001627, + "MSE/layer0": 732.1211085001627, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.22, + "input_norm": 31.997947177886957, + "input_norm/layer0": 31.997947177886957, + "learning_rate": 0.0004131578947368421, + "loss": 2.1053, + "max_norm": 51.30076217651367, + "max_norm/layer0": 51.30076217651367, + "mean_norm": 44.650190353393555, + "mean_norm/layer0": 44.650190353393555, + "multicode_k": 1, + "output_norm": 10.083865798314415, + "output_norm/layer0": 10.083865798314415, + "step": 2150 + }, + { + "MSE": 729.7699541219072, + "MSE/layer0": 729.7699541219072, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.23, + "input_norm": 31.997944199244184, + "input_norm/layer0": 31.997944199244184, + "learning_rate": 0.0004105263157894737, + "loss": 2.092, + "max_norm": 51.70292282104492, + "max_norm/layer0": 51.70292282104492, + "mean_norm": 44.99736022949219, + "mean_norm/layer0": 44.99736022949219, + "multicode_k": 1, + "output_norm": 10.171215546925865, + "output_norm/layer0": 10.171215546925865, + "step": 2200 + }, + { + "MSE": 727.7426215616864, + "MSE/layer0": 727.7426215616864, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.23, + "input_norm": 31.997949040730795, + "input_norm/layer0": 31.997949040730795, + "learning_rate": 0.00040789473684210524, + "loss": 2.0989, + "max_norm": 52.09043502807617, + "max_norm/layer0": 52.09043502807617, + "mean_norm": 45.34288787841797, + "mean_norm/layer0": 45.34288787841797, + "multicode_k": 1, + "output_norm": 10.245072917938227, + "output_norm/layer0": 10.245072917938227, + "step": 2250 + }, + { + "MSE": 725.7510225423177, + "MSE/layer0": 725.7510225423177, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.24, + "input_norm": 31.997945086161295, + "input_norm/layer0": 31.997945086161295, + "learning_rate": 0.00040526315789473684, + "loss": 2.0921, + "max_norm": 52.48381423950195, + "max_norm/layer0": 52.48381423950195, + "mean_norm": 45.685386657714844, + "mean_norm/layer0": 45.685386657714844, + "multicode_k": 1, + "output_norm": 10.316563812891642, + "output_norm/layer0": 10.316563812891642, + "step": 2300 + }, + { + "MSE": 723.730980834961, + "MSE/layer0": 723.730980834961, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.25, + "input_norm": 31.997938013076794, + "input_norm/layer0": 31.997938013076794, + "learning_rate": 0.00040263157894736843, + "loss": 2.0863, + "max_norm": 52.871910095214844, + "max_norm/layer0": 52.871910095214844, + "mean_norm": 46.027950286865234, + "mean_norm/layer0": 46.027950286865234, + "multicode_k": 1, + "output_norm": 10.396288099288938, + "output_norm/layer0": 10.396288099288938, + "step": 2350 + }, + { + "MSE": 721.850106608073, + "MSE/layer0": 721.850106608073, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.25, + "input_norm": 31.99794317245484, + "input_norm/layer0": 31.99794317245484, + "learning_rate": 0.0004, + "loss": 2.0883, + "max_norm": 53.25300598144531, + "max_norm/layer0": 53.25300598144531, + "mean_norm": 46.366220474243164, + "mean_norm/layer0": 46.366220474243164, + "multicode_k": 1, + "output_norm": 10.462737544377642, + "output_norm/layer0": 10.462737544377642, + "step": 2400 + }, + { + "MSE": 720.002911987305, + "MSE/layer0": 720.002911987305, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.26, + "input_norm": 31.997945442199722, + "input_norm/layer0": 31.997945442199722, + "learning_rate": 0.0003973684210526316, + "loss": 2.0813, + "max_norm": 53.6557502746582, + "max_norm/layer0": 53.6557502746582, + "mean_norm": 46.70218849182129, + "mean_norm/layer0": 46.70218849182129, + "multicode_k": 1, + "output_norm": 10.54251501719157, + "output_norm/layer0": 10.54251501719157, + "step": 2450 + }, + { + "MSE": 717.8726328531905, + "MSE/layer0": 717.8726328531905, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.26, + "input_norm": 31.997946141560867, + "input_norm/layer0": 31.997946141560867, + "learning_rate": 0.00039473684210526315, + "loss": 2.0847, + "max_norm": 54.013648986816406, + "max_norm/layer0": 54.013648986816406, + "mean_norm": 47.03492546081543, + "mean_norm/layer0": 47.03492546081543, + "multicode_k": 1, + "output_norm": 10.61746094703674, + "output_norm/layer0": 10.61746094703674, + "step": 2500 + }, + { + "epoch": 0.26, + "eval_MSE/layer0": 716.9390104187793, + "eval_accuracy": 0.5142129041984603, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997947305666536, + "eval_loss": 2.0790653228759766, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 10.657726783760687, + "eval_runtime": 73.6422, + "eval_samples_per_second": 62.776, + "eval_steps_per_second": 7.849, + "step": 2500 + }, + { + "MSE": 715.8716929117836, + "MSE/layer0": 715.8716929117836, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.27, + "input_norm": 31.99793814023335, + "input_norm/layer0": 31.99793814023335, + "learning_rate": 0.00039210526315789474, + "loss": 2.0789, + "max_norm": 54.395057678222656, + "max_norm/layer0": 54.395057678222656, + "mean_norm": 47.36547088623047, + "mean_norm/layer0": 47.36547088623047, + "multicode_k": 1, + "output_norm": 10.687965892155965, + "output_norm/layer0": 10.687965892155965, + "step": 2550 + }, + { + "MSE": 713.9484742228188, + "MSE/layer0": 713.9484742228188, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.27, + "input_norm": 31.997940645217888, + "input_norm/layer0": 31.997940645217888, + "learning_rate": 0.00038947368421052633, + "loss": 2.0747, + "max_norm": 54.81391525268555, + "max_norm/layer0": 54.81391525268555, + "mean_norm": 47.6934928894043, + "mean_norm/layer0": 47.6934928894043, + "multicode_k": 1, + "output_norm": 10.762619382540386, + "output_norm/layer0": 10.762619382540386, + "step": 2600 + }, + { + "MSE": 711.9854763793942, + "MSE/layer0": 711.9854763793942, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.28, + "input_norm": 31.997925097147615, + "input_norm/layer0": 31.997925097147615, + "learning_rate": 0.00038684210526315787, + "loss": 2.0679, + "max_norm": 55.21110916137695, + "max_norm/layer0": 55.21110916137695, + "mean_norm": 48.01936340332031, + "mean_norm/layer0": 48.01936340332031, + "multicode_k": 1, + "output_norm": 10.838534935315447, + "output_norm/layer0": 10.838534935315447, + "step": 2650 + }, + { + "MSE": 710.4415082804362, + "MSE/layer0": 710.4415082804362, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.28, + "input_norm": 31.997930930455517, + "input_norm/layer0": 31.997930930455517, + "learning_rate": 0.00038421052631578946, + "loss": 2.0619, + "max_norm": 55.63144302368164, + "max_norm/layer0": 55.63144302368164, + "mean_norm": 48.34212875366211, + "mean_norm/layer0": 48.34212875366211, + "multicode_k": 1, + "output_norm": 10.893479135831196, + "output_norm/layer0": 10.893479135831196, + "step": 2700 + }, + { + "MSE": 708.5378164672845, + "MSE/layer0": 708.5378164672845, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.29, + "input_norm": 31.99792820294698, + "input_norm/layer0": 31.99792820294698, + "learning_rate": 0.00038157894736842105, + "loss": 2.0461, + "max_norm": 56.01336669921875, + "max_norm/layer0": 56.01336669921875, + "mean_norm": 48.66323280334473, + "mean_norm/layer0": 48.66323280334473, + "multicode_k": 1, + "output_norm": 10.971131575902309, + "output_norm/layer0": 10.971131575902309, + "step": 2750 + }, + { + "MSE": 706.6155220540361, + "MSE/layer0": 706.6155220540361, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.29, + "input_norm": 31.997930272420245, + "input_norm/layer0": 31.997930272420245, + "learning_rate": 0.00037894736842105265, + "loss": 2.0594, + "max_norm": 56.40309143066406, + "max_norm/layer0": 56.40309143066406, + "mean_norm": 48.980411529541016, + "mean_norm/layer0": 48.980411529541016, + "multicode_k": 1, + "output_norm": 11.042961815198257, + "output_norm/layer0": 11.042961815198257, + "step": 2800 + }, + { + "MSE": 704.6534555053711, + "MSE/layer0": 704.6534555053711, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.3, + "input_norm": 31.99792085011799, + "input_norm/layer0": 31.99792085011799, + "learning_rate": 0.00037631578947368424, + "loss": 2.0499, + "max_norm": 56.79050064086914, + "max_norm/layer0": 56.79050064086914, + "mean_norm": 49.293588638305664, + "mean_norm/layer0": 49.293588638305664, + "multicode_k": 1, + "output_norm": 11.11463791847229, + "output_norm/layer0": 11.11463791847229, + "step": 2850 + }, + { + "MSE": 702.691480916341, + "MSE/layer0": 702.691480916341, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.3, + "input_norm": 31.997921177546203, + "input_norm/layer0": 31.997921177546203, + "learning_rate": 0.0003736842105263158, + "loss": 2.0472, + "max_norm": 57.16228103637695, + "max_norm/layer0": 57.16228103637695, + "mean_norm": 49.60378646850586, + "mean_norm/layer0": 49.60378646850586, + "multicode_k": 1, + "output_norm": 11.188902417818706, + "output_norm/layer0": 11.188902417818706, + "step": 2900 + }, + { + "MSE": 700.9804660034181, + "MSE/layer0": 700.9804660034181, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.31, + "input_norm": 31.997924566268914, + "input_norm/layer0": 31.997924566268914, + "learning_rate": 0.00037105263157894737, + "loss": 2.0557, + "max_norm": 57.52459716796875, + "max_norm/layer0": 57.52459716796875, + "mean_norm": 49.91103553771973, + "mean_norm/layer0": 49.91103553771973, + "multicode_k": 1, + "output_norm": 11.253552745183304, + "output_norm/layer0": 11.253552745183304, + "step": 2950 + }, + { + "MSE": 699.5130490112299, + "MSE/layer0": 699.5130490112299, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.31, + "input_norm": 31.997922519048053, + "input_norm/layer0": 31.997922519048053, + "learning_rate": 0.00036842105263157896, + "loss": 2.0439, + "max_norm": 57.87739562988281, + "max_norm/layer0": 57.87739562988281, + "mean_norm": 50.21486854553223, + "mean_norm/layer0": 50.21486854553223, + "multicode_k": 1, + "output_norm": 11.316310184796652, + "output_norm/layer0": 11.316310184796652, + "step": 3000 + }, + { + "epoch": 0.31, + "eval_MSE/layer0": 698.7265792011616, + "eval_accuracy": 0.5184875063671823, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997919895214224, + "eval_loss": 2.0482470989227295, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 11.359921689315088, + "eval_runtime": 74.2109, + "eval_samples_per_second": 62.295, + "eval_steps_per_second": 7.789, + "step": 3000 + }, + { + "MSE": 697.8663801066077, + "MSE/layer0": 697.8663801066077, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.32, + "input_norm": 31.997911771138504, + "input_norm/layer0": 31.997911771138504, + "learning_rate": 0.00036578947368421055, + "loss": 2.0511, + "max_norm": 58.24200439453125, + "max_norm/layer0": 58.24200439453125, + "mean_norm": 50.51446723937988, + "mean_norm/layer0": 50.51446723937988, + "multicode_k": 1, + "output_norm": 11.388139980634046, + "output_norm/layer0": 11.388139980634046, + "step": 3050 + }, + { + "MSE": 696.0450835164395, + "MSE/layer0": 696.0450835164395, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.32, + "input_norm": 31.9979209582011, + "input_norm/layer0": 31.9979209582011, + "learning_rate": 0.00036315789473684214, + "loss": 2.0466, + "max_norm": 58.58406066894531, + "max_norm/layer0": 58.58406066894531, + "mean_norm": 50.81120681762695, + "mean_norm/layer0": 50.81120681762695, + "multicode_k": 1, + "output_norm": 11.455551563898727, + "output_norm/layer0": 11.455551563898727, + "step": 3100 + }, + { + "MSE": 694.5301999918622, + "MSE/layer0": 694.5301999918622, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.33, + "input_norm": 31.99790574709574, + "input_norm/layer0": 31.99790574709574, + "learning_rate": 0.0003605263157894737, + "loss": 2.0294, + "max_norm": 58.931087493896484, + "max_norm/layer0": 58.931087493896484, + "mean_norm": 51.104164123535156, + "mean_norm/layer0": 51.104164123535156, + "multicode_k": 1, + "output_norm": 11.512675134340917, + "output_norm/layer0": 11.512675134340917, + "step": 3150 + }, + { + "MSE": 692.5095411173497, + "MSE/layer0": 692.5095411173497, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.33, + "input_norm": 31.997909634908044, + "input_norm/layer0": 31.997909634908044, + "learning_rate": 0.0003578947368421053, + "loss": 2.0455, + "max_norm": 59.2867546081543, + "max_norm/layer0": 59.2867546081543, + "mean_norm": 51.39415168762207, + "mean_norm/layer0": 51.39415168762207, + "multicode_k": 1, + "output_norm": 11.587491785685224, + "output_norm/layer0": 11.587491785685224, + "step": 3200 + }, + { + "MSE": 691.1425885009767, + "MSE/layer0": 691.1425885009767, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.34, + "input_norm": 31.99791768709818, + "input_norm/layer0": 31.99791768709818, + "learning_rate": 0.00035526315789473687, + "loss": 2.0348, + "max_norm": 59.64825439453125, + "max_norm/layer0": 59.64825439453125, + "mean_norm": 51.68109130859375, + "mean_norm/layer0": 51.68109130859375, + "multicode_k": 1, + "output_norm": 11.643148959477745, + "output_norm/layer0": 11.643148959477745, + "step": 3250 + }, + { + "MSE": 689.2906094360355, + "MSE/layer0": 689.2906094360355, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.34, + "input_norm": 31.997913980483997, + "input_norm/layer0": 31.997913980483997, + "learning_rate": 0.0003526315789473684, + "loss": 2.0293, + "max_norm": 59.97624206542969, + "max_norm/layer0": 59.97624206542969, + "mean_norm": 51.965484619140625, + "mean_norm/layer0": 51.965484619140625, + "multicode_k": 1, + "output_norm": 11.714975148836775, + "output_norm/layer0": 11.714975148836775, + "step": 3300 + }, + { + "MSE": 688.0525922648112, + "MSE/layer0": 688.0525922648112, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.35, + "input_norm": 31.997908350626624, + "input_norm/layer0": 31.997908350626624, + "learning_rate": 0.00035, + "loss": 2.0389, + "max_norm": 60.30556869506836, + "max_norm/layer0": 60.30556869506836, + "mean_norm": 52.2458438873291, + "mean_norm/layer0": 52.2458438873291, + "multicode_k": 1, + "output_norm": 11.772027517954506, + "output_norm/layer0": 11.772027517954506, + "step": 3350 + }, + { + "MSE": 686.4814953613279, + "MSE/layer0": 686.4814953613279, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.35, + "input_norm": 31.997902415593472, + "input_norm/layer0": 31.997902415593472, + "learning_rate": 0.0003473684210526316, + "loss": 2.0266, + "max_norm": 60.628334045410156, + "max_norm/layer0": 60.628334045410156, + "mean_norm": 52.522024154663086, + "mean_norm/layer0": 52.522024154663086, + "multicode_k": 1, + "output_norm": 11.842156640688584, + "output_norm/layer0": 11.842156640688584, + "step": 3400 + }, + { + "MSE": 684.6515231323242, + "MSE/layer0": 684.6515231323242, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.36, + "input_norm": 31.99791290283203, + "input_norm/layer0": 31.99791290283203, + "learning_rate": 0.0003447368421052632, + "loss": 2.0248, + "max_norm": 60.95072555541992, + "max_norm/layer0": 60.95072555541992, + "mean_norm": 52.79400825500488, + "mean_norm/layer0": 52.79400825500488, + "multicode_k": 1, + "output_norm": 11.907371897697445, + "output_norm/layer0": 11.907371897697445, + "step": 3450 + }, + { + "MSE": 683.5430062866212, + "MSE/layer0": 683.5430062866212, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.37, + "input_norm": 31.9979167175293, + "input_norm/layer0": 31.9979167175293, + "learning_rate": 0.00034210526315789477, + "loss": 2.0263, + "max_norm": 61.270816802978516, + "max_norm/layer0": 61.270816802978516, + "mean_norm": 53.06429481506348, + "mean_norm/layer0": 53.06429481506348, + "multicode_k": 1, + "output_norm": 11.956860675811765, + "output_norm/layer0": 11.956860675811765, + "step": 3500 + }, + { + "epoch": 0.37, + "eval_MSE/layer0": 682.2680427869782, + "eval_accuracy": 0.5224062440993215, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997916449774355, + "eval_loss": 2.0253396034240723, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 12.010468493388789, + "eval_runtime": 73.2716, + "eval_samples_per_second": 63.094, + "eval_steps_per_second": 7.888, + "step": 3500 + }, + { + "MSE": 682.0599540201822, + "MSE/layer0": 682.0599540201822, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.37, + "input_norm": 31.9979091612498, + "input_norm/layer0": 31.9979091612498, + "learning_rate": 0.0003394736842105263, + "loss": 2.035, + "max_norm": 61.60363006591797, + "max_norm/layer0": 61.60363006591797, + "mean_norm": 53.33056831359863, + "mean_norm/layer0": 53.33056831359863, + "multicode_k": 1, + "output_norm": 12.018342121442167, + "output_norm/layer0": 12.018342121442167, + "step": 3550 + }, + { + "MSE": 680.5750654093424, + "MSE/layer0": 680.5750654093424, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.38, + "input_norm": 31.997909587224328, + "input_norm/layer0": 31.997909587224328, + "learning_rate": 0.0003368421052631579, + "loss": 2.0232, + "max_norm": 61.922420501708984, + "max_norm/layer0": 61.922420501708984, + "mean_norm": 53.59366035461426, + "mean_norm/layer0": 53.59366035461426, + "multicode_k": 1, + "output_norm": 12.078021968205773, + "output_norm/layer0": 12.078021968205773, + "step": 3600 + }, + { + "MSE": 678.8478289794925, + "MSE/layer0": 678.8478289794925, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.38, + "input_norm": 31.99789888381958, + "input_norm/layer0": 31.99789888381958, + "learning_rate": 0.00033421052631578944, + "loss": 2.0171, + "max_norm": 62.24449157714844, + "max_norm/layer0": 62.24449157714844, + "mean_norm": 53.85357475280762, + "mean_norm/layer0": 53.85357475280762, + "multicode_k": 1, + "output_norm": 12.149001522064214, + "output_norm/layer0": 12.149001522064214, + "step": 3650 + }, + { + "MSE": 677.7631386311848, + "MSE/layer0": 677.7631386311848, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.39, + "input_norm": 31.997902571360274, + "input_norm/layer0": 31.997902571360274, + "learning_rate": 0.00033157894736842103, + "loss": 2.0212, + "max_norm": 62.564937591552734, + "max_norm/layer0": 62.564937591552734, + "mean_norm": 54.10923385620117, + "mean_norm/layer0": 54.10923385620117, + "multicode_k": 1, + "output_norm": 12.200160818099977, + "output_norm/layer0": 12.200160818099977, + "step": 3700 + }, + { + "MSE": 676.4079176839191, + "MSE/layer0": 676.4079176839191, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.39, + "input_norm": 31.99789404869079, + "input_norm/layer0": 31.99789404869079, + "learning_rate": 0.0003289473684210527, + "loss": 2.0184, + "max_norm": 62.88063430786133, + "max_norm/layer0": 62.88063430786133, + "mean_norm": 54.362863540649414, + "mean_norm/layer0": 54.362863540649414, + "multicode_k": 1, + "output_norm": 12.259288868904115, + "output_norm/layer0": 12.259288868904115, + "step": 3750 + }, + { + "MSE": 675.2395422363282, + "MSE/layer0": 675.2395422363282, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.4, + "input_norm": 31.99789286295573, + "input_norm/layer0": 31.99789286295573, + "learning_rate": 0.0003263157894736842, + "loss": 2.0058, + "max_norm": 63.18323516845703, + "max_norm/layer0": 63.18323516845703, + "mean_norm": 54.61160659790039, + "mean_norm/layer0": 54.61160659790039, + "multicode_k": 1, + "output_norm": 12.305311093330385, + "output_norm/layer0": 12.305311093330385, + "step": 3800 + }, + { + "MSE": 673.5289611816404, + "MSE/layer0": 673.5289611816404, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.4, + "input_norm": 31.997895905176787, + "input_norm/layer0": 31.997895905176787, + "learning_rate": 0.0003236842105263158, + "loss": 2.0147, + "max_norm": 63.47829055786133, + "max_norm/layer0": 63.47829055786133, + "mean_norm": 54.85733413696289, + "mean_norm/layer0": 54.85733413696289, + "multicode_k": 1, + "output_norm": 12.368880640665692, + "output_norm/layer0": 12.368880640665692, + "step": 3850 + }, + { + "MSE": 672.7262348429363, + "MSE/layer0": 672.7262348429363, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.41, + "input_norm": 31.997892808914187, + "input_norm/layer0": 31.997892808914187, + "learning_rate": 0.0003210526315789474, + "loss": 2.0011, + "max_norm": 63.7920036315918, + "max_norm/layer0": 63.7920036315918, + "mean_norm": 55.099992752075195, + "mean_norm/layer0": 55.099992752075195, + "multicode_k": 1, + "output_norm": 12.413625540733335, + "output_norm/layer0": 12.413625540733335, + "step": 3900 + }, + { + "MSE": 671.2364042154949, + "MSE/layer0": 671.2364042154949, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.41, + "input_norm": 31.997892484664916, + "input_norm/layer0": 31.997892484664916, + "learning_rate": 0.00031842105263157894, + "loss": 2.0068, + "max_norm": 64.07488250732422, + "max_norm/layer0": 64.07488250732422, + "mean_norm": 55.33942985534668, + "mean_norm/layer0": 55.33942985534668, + "multicode_k": 1, + "output_norm": 12.478335504531861, + "output_norm/layer0": 12.478335504531861, + "step": 3950 + }, + { + "MSE": 669.9738427734378, + "MSE/layer0": 669.9738427734378, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.42, + "input_norm": 31.997889916102086, + "input_norm/layer0": 31.997889916102086, + "learning_rate": 0.00031578947368421053, + "loss": 1.9906, + "max_norm": 64.34879302978516, + "max_norm/layer0": 64.34879302978516, + "mean_norm": 55.576541900634766, + "mean_norm/layer0": 55.576541900634766, + "multicode_k": 1, + "output_norm": 12.524646544456482, + "output_norm/layer0": 12.524646544456482, + "step": 4000 + }, + { + "epoch": 0.42, + "eval_MSE/layer0": 669.1965223770751, + "eval_accuracy": 0.5253332978103237, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997898890449704, + "eval_loss": 2.006638526916504, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 12.556819209953474, + "eval_runtime": 73.3751, + "eval_samples_per_second": 63.005, + "eval_steps_per_second": 7.877, + "step": 4000 + }, + { + "MSE": 668.3390091959637, + "MSE/layer0": 668.3390091959637, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.42, + "input_norm": 31.99788640658062, + "input_norm/layer0": 31.99788640658062, + "learning_rate": 0.00031315789473684207, + "loss": 1.9962, + "max_norm": 64.65262603759766, + "max_norm/layer0": 64.65262603759766, + "mean_norm": 55.811140060424805, + "mean_norm/layer0": 55.811140060424805, + "multicode_k": 1, + "output_norm": 12.584023051261894, + "output_norm/layer0": 12.584023051261894, + "step": 4050 + }, + { + "MSE": 667.4144735717773, + "MSE/layer0": 667.4144735717773, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.43, + "input_norm": 31.99788414637247, + "input_norm/layer0": 31.99788414637247, + "learning_rate": 0.0003105263157894737, + "loss": 2.0038, + "max_norm": 64.9332275390625, + "max_norm/layer0": 64.9332275390625, + "mean_norm": 56.04119682312012, + "mean_norm/layer0": 56.04119682312012, + "multicode_k": 1, + "output_norm": 12.633416105906175, + "output_norm/layer0": 12.633416105906175, + "step": 4100 + }, + { + "MSE": 666.502211812337, + "MSE/layer0": 666.502211812337, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.43, + "input_norm": 31.997885338465373, + "input_norm/layer0": 31.997885338465373, + "learning_rate": 0.0003078947368421053, + "loss": 2.0046, + "max_norm": 65.20265197753906, + "max_norm/layer0": 65.20265197753906, + "mean_norm": 56.26777458190918, + "mean_norm/layer0": 56.26777458190918, + "multicode_k": 1, + "output_norm": 12.67455391089122, + "output_norm/layer0": 12.67455391089122, + "step": 4150 + }, + { + "MSE": 665.0832258097332, + "MSE/layer0": 665.0832258097332, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.44, + "input_norm": 31.997875661849967, + "input_norm/layer0": 31.997875661849967, + "learning_rate": 0.00030526315789473684, + "loss": 2.0066, + "max_norm": 65.46887969970703, + "max_norm/layer0": 65.46887969970703, + "mean_norm": 56.49208450317383, + "mean_norm/layer0": 56.49208450317383, + "multicode_k": 1, + "output_norm": 12.73067569255829, + "output_norm/layer0": 12.73067569255829, + "step": 4200 + }, + { + "MSE": 663.9124774169925, + "MSE/layer0": 663.9124774169925, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.44, + "input_norm": 31.997874129613244, + "input_norm/layer0": 31.997874129613244, + "learning_rate": 0.00030263157894736844, + "loss": 2.0006, + "max_norm": 65.73078918457031, + "max_norm/layer0": 65.73078918457031, + "mean_norm": 56.712989807128906, + "mean_norm/layer0": 56.712989807128906, + "multicode_k": 1, + "output_norm": 12.783326719601945, + "output_norm/layer0": 12.783326719601945, + "step": 4250 + }, + { + "MSE": 663.0191631062823, + "MSE/layer0": 663.0191631062823, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.45, + "input_norm": 31.99787082672119, + "input_norm/layer0": 31.99787082672119, + "learning_rate": 0.0003, + "loss": 1.9862, + "max_norm": 65.99481964111328, + "max_norm/layer0": 65.99481964111328, + "mean_norm": 56.93141746520996, + "mean_norm/layer0": 56.93141746520996, + "multicode_k": 1, + "output_norm": 12.824343484242757, + "output_norm/layer0": 12.824343484242757, + "step": 4300 + }, + { + "MSE": 661.9175501505531, + "MSE/layer0": 661.9175501505531, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.45, + "input_norm": 31.997863101959226, + "input_norm/layer0": 31.997863101959226, + "learning_rate": 0.00029736842105263157, + "loss": 1.9891, + "max_norm": 66.25289916992188, + "max_norm/layer0": 66.25289916992188, + "mean_norm": 57.14705848693848, + "mean_norm/layer0": 57.14705848693848, + "multicode_k": 1, + "output_norm": 12.873331023852028, + "output_norm/layer0": 12.873331023852028, + "step": 4350 + }, + { + "MSE": 660.8278486124677, + "MSE/layer0": 660.8278486124677, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.46, + "input_norm": 31.99786113739014, + "input_norm/layer0": 31.99786113739014, + "learning_rate": 0.00029473684210526316, + "loss": 1.9874, + "max_norm": 66.49950408935547, + "max_norm/layer0": 66.49950408935547, + "mean_norm": 57.3592414855957, + "mean_norm/layer0": 57.3592414855957, + "multicode_k": 1, + "output_norm": 12.925755645434062, + "output_norm/layer0": 12.925755645434062, + "step": 4400 + }, + { + "MSE": 659.7812182617188, + "MSE/layer0": 659.7812182617188, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.46, + "input_norm": 31.997859818140668, + "input_norm/layer0": 31.997859818140668, + "learning_rate": 0.00029210526315789475, + "loss": 1.9894, + "max_norm": 66.74647521972656, + "max_norm/layer0": 66.74647521972656, + "mean_norm": 57.56860542297363, + "mean_norm/layer0": 57.56860542297363, + "multicode_k": 1, + "output_norm": 12.969949612617494, + "output_norm/layer0": 12.969949612617494, + "step": 4450 + }, + { + "MSE": 658.2862462361654, + "MSE/layer0": 658.2862462361654, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.47, + "input_norm": 31.997855517069482, + "input_norm/layer0": 31.997855517069482, + "learning_rate": 0.00028947368421052634, + "loss": 1.9852, + "max_norm": 67.0057373046875, + "max_norm/layer0": 67.0057373046875, + "mean_norm": 57.77582931518555, + "mean_norm/layer0": 57.77582931518555, + "multicode_k": 1, + "output_norm": 13.019407332738238, + "output_norm/layer0": 13.019407332738238, + "step": 4500 + }, + { + "epoch": 0.47, + "eval_MSE/layer0": 657.5871718611108, + "eval_accuracy": 0.5279040641917702, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997854675071842, + "eval_loss": 1.9898165464401245, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 13.052642994207561, + "eval_runtime": 74.0479, + "eval_samples_per_second": 62.433, + "eval_steps_per_second": 7.806, + "step": 4500 + }, + { + "MSE": 657.2974259440105, + "MSE/layer0": 657.2974259440105, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.48, + "input_norm": 31.99785115559897, + "input_norm/layer0": 31.99785115559897, + "learning_rate": 0.0002868421052631579, + "loss": 1.9727, + "max_norm": 67.25566101074219, + "max_norm/layer0": 67.25566101074219, + "mean_norm": 57.98077964782715, + "mean_norm/layer0": 57.98077964782715, + "multicode_k": 1, + "output_norm": 13.063522001902262, + "output_norm/layer0": 13.063522001902262, + "step": 4550 + }, + { + "MSE": 656.5759895833334, + "MSE/layer0": 656.5759895833334, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.48, + "input_norm": 31.997858088811242, + "input_norm/layer0": 31.997858088811242, + "learning_rate": 0.00028421052631578947, + "loss": 1.9897, + "max_norm": 67.49605560302734, + "max_norm/layer0": 67.49605560302734, + "mean_norm": 58.182559967041016, + "mean_norm/layer0": 58.182559967041016, + "multicode_k": 1, + "output_norm": 13.099744346936546, + "output_norm/layer0": 13.099744346936546, + "step": 4600 + }, + { + "MSE": 655.8373800659178, + "MSE/layer0": 655.8373800659178, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.49, + "input_norm": 31.997857850392663, + "input_norm/layer0": 31.997857850392663, + "learning_rate": 0.00028157894736842106, + "loss": 1.9918, + "max_norm": 67.72962188720703, + "max_norm/layer0": 67.72962188720703, + "mean_norm": 58.38115119934082, + "mean_norm/layer0": 58.38115119934082, + "multicode_k": 1, + "output_norm": 13.13247790972392, + "output_norm/layer0": 13.13247790972392, + "step": 4650 + }, + { + "MSE": 654.6057424926755, + "MSE/layer0": 654.6057424926755, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.49, + "input_norm": 31.997855739593504, + "input_norm/layer0": 31.997855739593504, + "learning_rate": 0.0002789473684210526, + "loss": 1.9908, + "max_norm": 67.96855163574219, + "max_norm/layer0": 67.96855163574219, + "mean_norm": 58.57722091674805, + "mean_norm/layer0": 58.57722091674805, + "multicode_k": 1, + "output_norm": 13.187800091107682, + "output_norm/layer0": 13.187800091107682, + "step": 4700 + }, + { + "MSE": 653.7336292521161, + "MSE/layer0": 653.7336292521161, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.5, + "input_norm": 31.997861677805588, + "input_norm/layer0": 31.997861677805588, + "learning_rate": 0.00027631578947368425, + "loss": 1.9919, + "max_norm": 68.20356750488281, + "max_norm/layer0": 68.20356750488281, + "mean_norm": 58.77041053771973, + "mean_norm/layer0": 58.77041053771973, + "multicode_k": 1, + "output_norm": 13.224705770810434, + "output_norm/layer0": 13.224705770810434, + "step": 4750 + }, + { + "MSE": 652.4711893717447, + "MSE/layer0": 652.4711893717447, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.5, + "input_norm": 31.997852430343634, + "input_norm/layer0": 31.997852430343634, + "learning_rate": 0.00027368421052631584, + "loss": 1.9777, + "max_norm": 68.42557525634766, + "max_norm/layer0": 68.42557525634766, + "mean_norm": 58.96235466003418, + "mean_norm/layer0": 58.96235466003418, + "multicode_k": 1, + "output_norm": 13.275700616836549, + "output_norm/layer0": 13.275700616836549, + "step": 4800 + }, + { + "MSE": 651.660216674805, + "MSE/layer0": 651.660216674805, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.51, + "input_norm": 31.997857831319166, + "input_norm/layer0": 31.997857831319166, + "learning_rate": 0.0002710526315789474, + "loss": 1.9728, + "max_norm": 68.6562271118164, + "max_norm/layer0": 68.6562271118164, + "mean_norm": 59.151214599609375, + "mean_norm/layer0": 59.151214599609375, + "multicode_k": 1, + "output_norm": 13.316913062731425, + "output_norm/layer0": 13.316913062731425, + "step": 4850 + }, + { + "MSE": 651.1180463663741, + "MSE/layer0": 651.1180463663741, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.51, + "input_norm": 31.997851276397704, + "input_norm/layer0": 31.997851276397704, + "learning_rate": 0.00026842105263157897, + "loss": 1.9806, + "max_norm": 68.8842544555664, + "max_norm/layer0": 68.8842544555664, + "mean_norm": 59.336891174316406, + "mean_norm/layer0": 59.336891174316406, + "multicode_k": 1, + "output_norm": 13.348248120943708, + "output_norm/layer0": 13.348248120943708, + "step": 4900 + }, + { + "MSE": 650.0774853515621, + "MSE/layer0": 650.0774853515621, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.52, + "input_norm": 31.997846142450957, + "input_norm/layer0": 31.997846142450957, + "learning_rate": 0.0002657894736842105, + "loss": 1.9718, + "max_norm": 69.09481811523438, + "max_norm/layer0": 69.09481811523438, + "mean_norm": 59.52014923095703, + "mean_norm/layer0": 59.52014923095703, + "multicode_k": 1, + "output_norm": 13.38570951779683, + "output_norm/layer0": 13.38570951779683, + "step": 4950 + }, + { + "MSE": 649.2541728719073, + "MSE/layer0": 649.2541728719073, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.52, + "input_norm": 31.997852964401247, + "input_norm/layer0": 31.997852964401247, + "learning_rate": 0.0002631578947368421, + "loss": 1.9687, + "max_norm": 69.3100357055664, + "max_norm/layer0": 69.3100357055664, + "mean_norm": 59.70068359375, + "mean_norm/layer0": 59.70068359375, + "multicode_k": 1, + "output_norm": 13.423000381787617, + "output_norm/layer0": 13.423000381787617, + "step": 5000 + }, + { + "epoch": 0.52, + "eval_MSE/layer0": 648.246248562512, + "eval_accuracy": 0.5299863891896716, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997853133679993, + "eval_loss": 1.975706934928894, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 13.449585199510798, + "eval_runtime": 73.7352, + "eval_samples_per_second": 62.697, + "eval_steps_per_second": 7.839, + "step": 5000 + }, + { + "MSE": 648.4500269571938, + "MSE/layer0": 648.4500269571938, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.53, + "input_norm": 31.99784724235535, + "input_norm/layer0": 31.99784724235535, + "learning_rate": 0.0002605263157894737, + "loss": 1.9816, + "max_norm": 69.5140151977539, + "max_norm/layer0": 69.5140151977539, + "mean_norm": 59.87860107421875, + "mean_norm/layer0": 59.87860107421875, + "multicode_k": 1, + "output_norm": 13.459953915278113, + "output_norm/layer0": 13.459953915278113, + "step": 5050 + }, + { + "MSE": 647.5120207722985, + "MSE/layer0": 647.5120207722985, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.53, + "input_norm": 31.997845083872484, + "input_norm/layer0": 31.997845083872484, + "learning_rate": 0.0002578947368421053, + "loss": 1.9778, + "max_norm": 69.72222137451172, + "max_norm/layer0": 69.72222137451172, + "mean_norm": 60.054636001586914, + "mean_norm/layer0": 60.054636001586914, + "multicode_k": 1, + "output_norm": 13.495457221666976, + "output_norm/layer0": 13.495457221666976, + "step": 5100 + }, + { + "MSE": 646.8909526570638, + "MSE/layer0": 646.8909526570638, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.54, + "input_norm": 31.99783927281696, + "input_norm/layer0": 31.99783927281696, + "learning_rate": 0.0002552631578947369, + "loss": 1.9608, + "max_norm": 69.93621826171875, + "max_norm/layer0": 69.93621826171875, + "mean_norm": 60.228532791137695, + "mean_norm/layer0": 60.228532791137695, + "multicode_k": 1, + "output_norm": 13.523821023305253, + "output_norm/layer0": 13.523821023305253, + "step": 5150 + }, + { + "MSE": 645.6001059977214, + "MSE/layer0": 645.6001059977214, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.54, + "input_norm": 31.997829329172767, + "input_norm/layer0": 31.997829329172767, + "learning_rate": 0.0002526315789473684, + "loss": 1.9514, + "max_norm": 70.1629867553711, + "max_norm/layer0": 70.1629867553711, + "mean_norm": 60.39993667602539, + "mean_norm/layer0": 60.39993667602539, + "multicode_k": 1, + "output_norm": 13.575601536432904, + "output_norm/layer0": 13.575601536432904, + "step": 5200 + }, + { + "MSE": 645.0477313232423, + "MSE/layer0": 645.0477313232423, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.55, + "input_norm": 31.997829844156904, + "input_norm/layer0": 31.997829844156904, + "learning_rate": 0.00025, + "loss": 1.953, + "max_norm": 70.36659240722656, + "max_norm/layer0": 70.36659240722656, + "mean_norm": 60.568695068359375, + "mean_norm/layer0": 60.568695068359375, + "multicode_k": 1, + "output_norm": 13.606370126406352, + "output_norm/layer0": 13.606370126406352, + "step": 5250 + }, + { + "MSE": 644.0795441691082, + "MSE/layer0": 644.0795441691082, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.55, + "input_norm": 31.997827720642086, + "input_norm/layer0": 31.997827720642086, + "learning_rate": 0.0002473684210526316, + "loss": 1.9664, + "max_norm": 70.58203125, + "max_norm/layer0": 70.58203125, + "mean_norm": 60.73503303527832, + "mean_norm/layer0": 60.73503303527832, + "multicode_k": 1, + "output_norm": 13.644356350898736, + "output_norm/layer0": 13.644356350898736, + "step": 5300 + }, + { + "MSE": 643.4398297119142, + "MSE/layer0": 643.4398297119142, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.56, + "input_norm": 31.99783255259196, + "input_norm/layer0": 31.99783255259196, + "learning_rate": 0.0002447368421052632, + "loss": 1.9612, + "max_norm": 70.80116271972656, + "max_norm/layer0": 70.80116271972656, + "mean_norm": 60.89903450012207, + "mean_norm/layer0": 60.89903450012207, + "multicode_k": 1, + "output_norm": 13.676611545880633, + "output_norm/layer0": 13.676611545880633, + "step": 5350 + }, + { + "MSE": 642.6565199788413, + "MSE/layer0": 642.6565199788413, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.56, + "input_norm": 31.997826932271334, + "input_norm/layer0": 31.997826932271334, + "learning_rate": 0.00024210526315789475, + "loss": 1.9695, + "max_norm": 71.0198745727539, + "max_norm/layer0": 71.0198745727539, + "mean_norm": 61.06051063537598, + "mean_norm/layer0": 61.06051063537598, + "multicode_k": 1, + "output_norm": 13.705395914713542, + "output_norm/layer0": 13.705395914713542, + "step": 5400 + }, + { + "MSE": 641.5518863932293, + "MSE/layer0": 641.5518863932293, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.57, + "input_norm": 31.99782320658366, + "input_norm/layer0": 31.99782320658366, + "learning_rate": 0.00023947368421052632, + "loss": 1.9708, + "max_norm": 71.22209930419922, + "max_norm/layer0": 71.22209930419922, + "mean_norm": 61.22001647949219, + "mean_norm/layer0": 61.22001647949219, + "multicode_k": 1, + "output_norm": 13.747722525596622, + "output_norm/layer0": 13.747722525596622, + "step": 5450 + }, + { + "MSE": 641.0277577718095, + "MSE/layer0": 641.0277577718095, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.57, + "input_norm": 31.997817249298095, + "input_norm/layer0": 31.997817249298095, + "learning_rate": 0.00023684210526315788, + "loss": 1.9672, + "max_norm": 71.42549896240234, + "max_norm/layer0": 71.42549896240234, + "mean_norm": 61.377342224121094, + "mean_norm/layer0": 61.377342224121094, + "multicode_k": 1, + "output_norm": 13.775313488642375, + "output_norm/layer0": 13.775313488642375, + "step": 5500 + }, + { + "epoch": 0.57, + "eval_MSE/layer0": 640.0821653411886, + "eval_accuracy": 0.5321348969378108, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997811444105338, + "eval_loss": 1.9619895219802856, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 13.80778875099279, + "eval_runtime": 73.8101, + "eval_samples_per_second": 62.634, + "eval_steps_per_second": 7.831, + "step": 5500 + }, + { + "MSE": 640.2260070800783, + "MSE/layer0": 640.2260070800783, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.58, + "input_norm": 31.997807060877484, + "input_norm/layer0": 31.997807060877484, + "learning_rate": 0.00023421052631578948, + "loss": 1.9526, + "max_norm": 71.6324691772461, + "max_norm/layer0": 71.6324691772461, + "mean_norm": 61.532691955566406, + "mean_norm/layer0": 61.532691955566406, + "multicode_k": 1, + "output_norm": 13.81434581597646, + "output_norm/layer0": 13.81434581597646, + "step": 5550 + }, + { + "MSE": 639.6603690592448, + "MSE/layer0": 639.6603690592448, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.58, + "input_norm": 31.997815796534205, + "input_norm/layer0": 31.997815796534205, + "learning_rate": 0.00023157894736842107, + "loss": 1.9592, + "max_norm": 71.83050537109375, + "max_norm/layer0": 71.83050537109375, + "mean_norm": 61.68556213378906, + "mean_norm/layer0": 61.68556213378906, + "multicode_k": 1, + "output_norm": 13.843803273836771, + "output_norm/layer0": 13.843803273836771, + "step": 5600 + }, + { + "MSE": 638.8630006917316, + "MSE/layer0": 638.8630006917316, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.59, + "input_norm": 31.997804651260374, + "input_norm/layer0": 31.997804651260374, + "learning_rate": 0.00022894736842105263, + "loss": 1.9582, + "max_norm": 72.0186767578125, + "max_norm/layer0": 72.0186767578125, + "mean_norm": 61.836381912231445, + "mean_norm/layer0": 61.836381912231445, + "multicode_k": 1, + "output_norm": 13.87206829547882, + "output_norm/layer0": 13.87206829547882, + "step": 5650 + }, + { + "MSE": 638.6114538574218, + "MSE/layer0": 638.6114538574218, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.6, + "input_norm": 31.997799615859993, + "input_norm/layer0": 31.997799615859993, + "learning_rate": 0.00022631578947368422, + "loss": 1.9581, + "max_norm": 72.212158203125, + "max_norm/layer0": 72.212158203125, + "mean_norm": 61.984375, + "mean_norm/layer0": 61.984375, + "multicode_k": 1, + "output_norm": 13.890618721644087, + "output_norm/layer0": 13.890618721644087, + "step": 5700 + }, + { + "MSE": 637.4200433349613, + "MSE/layer0": 637.4200433349613, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.6, + "input_norm": 31.9977961031596, + "input_norm/layer0": 31.9977961031596, + "learning_rate": 0.0002236842105263158, + "loss": 1.9563, + "max_norm": 72.40010833740234, + "max_norm/layer0": 72.40010833740234, + "mean_norm": 62.13043212890625, + "mean_norm/layer0": 62.13043212890625, + "multicode_k": 1, + "output_norm": 13.935336654980983, + "output_norm/layer0": 13.935336654980983, + "step": 5750 + }, + { + "MSE": 636.9881141153974, + "MSE/layer0": 636.9881141153974, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.61, + "input_norm": 31.997795972824097, + "input_norm/layer0": 31.997795972824097, + "learning_rate": 0.00022105263157894735, + "loss": 1.9652, + "max_norm": 72.58822631835938, + "max_norm/layer0": 72.58822631835938, + "mean_norm": 62.274553298950195, + "mean_norm/layer0": 62.274553298950195, + "multicode_k": 1, + "output_norm": 13.960987841288247, + "output_norm/layer0": 13.960987841288247, + "step": 5800 + }, + { + "MSE": 636.22215037028, + "MSE/layer0": 636.22215037028, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.61, + "input_norm": 31.997794774373368, + "input_norm/layer0": 31.997794774373368, + "learning_rate": 0.00021842105263157897, + "loss": 1.9509, + "max_norm": 72.77027130126953, + "max_norm/layer0": 72.77027130126953, + "mean_norm": 62.417043685913086, + "mean_norm/layer0": 62.417043685913086, + "multicode_k": 1, + "output_norm": 13.98557560602824, + "output_norm/layer0": 13.98557560602824, + "step": 5850 + }, + { + "MSE": 635.6220120239254, + "MSE/layer0": 635.6220120239254, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.62, + "input_norm": 31.997796500523897, + "input_norm/layer0": 31.997796500523897, + "learning_rate": 0.00021578947368421054, + "loss": 1.9637, + "max_norm": 72.93942260742188, + "max_norm/layer0": 72.93942260742188, + "mean_norm": 62.5573787689209, + "mean_norm/layer0": 62.5573787689209, + "multicode_k": 1, + "output_norm": 14.011822309494022, + "output_norm/layer0": 14.011822309494022, + "step": 5900 + }, + { + "MSE": 635.1990796915693, + "MSE/layer0": 635.1990796915693, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.62, + "input_norm": 31.99778350830077, + "input_norm/layer0": 31.99778350830077, + "learning_rate": 0.0002131578947368421, + "loss": 1.9417, + "max_norm": 73.11217498779297, + "max_norm/layer0": 73.11217498779297, + "mean_norm": 62.69554328918457, + "mean_norm/layer0": 62.69554328918457, + "multicode_k": 1, + "output_norm": 14.040199557940166, + "output_norm/layer0": 14.040199557940166, + "step": 5950 + }, + { + "MSE": 634.617561645508, + "MSE/layer0": 634.617561645508, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.63, + "input_norm": 31.9977766195933, + "input_norm/layer0": 31.9977766195933, + "learning_rate": 0.00021052631578947367, + "loss": 1.9441, + "max_norm": 73.27617645263672, + "max_norm/layer0": 73.27617645263672, + "mean_norm": 62.831491470336914, + "mean_norm/layer0": 62.831491470336914, + "multicode_k": 1, + "output_norm": 14.065582130750016, + "output_norm/layer0": 14.065582130750016, + "step": 6000 + }, + { + "epoch": 0.63, + "eval_MSE/layer0": 633.8831350106634, + "eval_accuracy": 0.5338761587531762, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997772803689244, + "eval_loss": 1.951315999031067, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 14.101806794000915, + "eval_runtime": 73.5977, + "eval_samples_per_second": 62.814, + "eval_steps_per_second": 7.854, + "step": 6000 + }, + { + "MSE": 633.8391249593099, + "MSE/layer0": 633.8391249593099, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.63, + "input_norm": 31.99777683258057, + "input_norm/layer0": 31.99777683258057, + "learning_rate": 0.0002078947368421053, + "loss": 1.9507, + "max_norm": 73.43240356445312, + "max_norm/layer0": 73.43240356445312, + "mean_norm": 62.96537971496582, + "mean_norm/layer0": 62.96537971496582, + "multicode_k": 1, + "output_norm": 14.0993266805013, + "output_norm/layer0": 14.0993266805013, + "step": 6050 + }, + { + "MSE": 633.1878758748373, + "MSE/layer0": 633.1878758748373, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.64, + "input_norm": 31.997768185933438, + "input_norm/layer0": 31.997768185933438, + "learning_rate": 0.00020526315789473685, + "loss": 1.9535, + "max_norm": 73.59780883789062, + "max_norm/layer0": 73.59780883789062, + "mean_norm": 63.09744453430176, + "mean_norm/layer0": 63.09744453430176, + "multicode_k": 1, + "output_norm": 14.12703340212504, + "output_norm/layer0": 14.12703340212504, + "step": 6100 + }, + { + "MSE": 632.4774736531577, + "MSE/layer0": 632.4774736531577, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.64, + "input_norm": 31.997762225468954, + "input_norm/layer0": 31.997762225468954, + "learning_rate": 0.00020263157894736842, + "loss": 1.9502, + "max_norm": 73.7634506225586, + "max_norm/layer0": 73.7634506225586, + "mean_norm": 63.227373123168945, + "mean_norm/layer0": 63.227373123168945, + "multicode_k": 1, + "output_norm": 14.155767776171366, + "output_norm/layer0": 14.155767776171366, + "step": 6150 + }, + { + "MSE": 632.0819724527997, + "MSE/layer0": 632.0819724527997, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.65, + "input_norm": 31.997758595148735, + "input_norm/layer0": 31.997758595148735, + "learning_rate": 0.0002, + "loss": 1.948, + "max_norm": 73.93152618408203, + "max_norm/layer0": 73.93152618408203, + "mean_norm": 63.35538673400879, + "mean_norm/layer0": 63.35538673400879, + "multicode_k": 1, + "output_norm": 14.17972202301026, + "output_norm/layer0": 14.17972202301026, + "step": 6200 + }, + { + "MSE": 631.3937511189779, + "MSE/layer0": 631.3937511189779, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.65, + "input_norm": 31.997760909398387, + "input_norm/layer0": 31.997760909398387, + "learning_rate": 0.00019736842105263157, + "loss": 1.9449, + "max_norm": 74.07744598388672, + "max_norm/layer0": 74.07744598388672, + "mean_norm": 63.481435775756836, + "mean_norm/layer0": 63.481435775756836, + "multicode_k": 1, + "output_norm": 14.207703741391498, + "output_norm/layer0": 14.207703741391498, + "step": 6250 + }, + { + "MSE": 631.1097898356121, + "MSE/layer0": 631.1097898356121, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.66, + "input_norm": 31.997752253214514, + "input_norm/layer0": 31.997752253214514, + "learning_rate": 0.00019473684210526317, + "loss": 1.9256, + "max_norm": 74.23124694824219, + "max_norm/layer0": 74.23124694824219, + "mean_norm": 63.605464935302734, + "mean_norm/layer0": 63.605464935302734, + "multicode_k": 1, + "output_norm": 14.22562705675761, + "output_norm/layer0": 14.22562705675761, + "step": 6300 + }, + { + "MSE": 630.4715811157231, + "MSE/layer0": 630.4715811157231, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.66, + "input_norm": 31.99775326093037, + "input_norm/layer0": 31.99775326093037, + "learning_rate": 0.00019210526315789473, + "loss": 1.9394, + "max_norm": 74.37789154052734, + "max_norm/layer0": 74.37789154052734, + "mean_norm": 63.72765922546387, + "mean_norm/layer0": 63.72765922546387, + "multicode_k": 1, + "output_norm": 14.252348532676702, + "output_norm/layer0": 14.252348532676702, + "step": 6350 + }, + { + "MSE": 629.5616383870444, + "MSE/layer0": 629.5616383870444, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.67, + "input_norm": 31.9977592086792, + "input_norm/layer0": 31.9977592086792, + "learning_rate": 0.00018947368421052632, + "loss": 1.948, + "max_norm": 74.52799987792969, + "max_norm/layer0": 74.52799987792969, + "mean_norm": 63.84817886352539, + "mean_norm/layer0": 63.84817886352539, + "multicode_k": 1, + "output_norm": 14.278619543711342, + "output_norm/layer0": 14.278619543711342, + "step": 6400 + }, + { + "MSE": 628.9405068969726, + "MSE/layer0": 628.9405068969726, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.67, + "input_norm": 31.9977388159434, + "input_norm/layer0": 31.9977388159434, + "learning_rate": 0.0001868421052631579, + "loss": 1.9365, + "max_norm": 74.66854095458984, + "max_norm/layer0": 74.66854095458984, + "mean_norm": 63.96674346923828, + "mean_norm/layer0": 63.96674346923828, + "multicode_k": 1, + "output_norm": 14.308290360768634, + "output_norm/layer0": 14.308290360768634, + "step": 6450 + }, + { + "MSE": 628.8358187866208, + "MSE/layer0": 628.8358187866208, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.68, + "input_norm": 31.997742996215806, + "input_norm/layer0": 31.997742996215806, + "learning_rate": 0.00018421052631578948, + "loss": 1.9408, + "max_norm": 74.80778503417969, + "max_norm/layer0": 74.80778503417969, + "mean_norm": 64.08341407775879, + "mean_norm/layer0": 64.08341407775879, + "multicode_k": 1, + "output_norm": 14.319794411659238, + "output_norm/layer0": 14.319794411659238, + "step": 6500 + }, + { + "epoch": 0.68, + "eval_MSE/layer0": 628.092910030562, + "eval_accuracy": 0.5357603583933366, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997742160687373, + "eval_loss": 1.9396723508834839, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 14.354976222496019, + "eval_runtime": 73.9338, + "eval_samples_per_second": 62.529, + "eval_steps_per_second": 7.818, + "step": 6500 + }, + { + "MSE": 628.4872816975908, + "MSE/layer0": 628.4872816975908, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.68, + "input_norm": 31.997741378148394, + "input_norm/layer0": 31.997741378148394, + "learning_rate": 0.00018157894736842107, + "loss": 1.9392, + "max_norm": 74.95079040527344, + "max_norm/layer0": 74.95079040527344, + "mean_norm": 64.19818496704102, + "mean_norm/layer0": 64.19818496704102, + "multicode_k": 1, + "output_norm": 14.340212704340617, + "output_norm/layer0": 14.340212704340617, + "step": 6550 + }, + { + "MSE": 627.595106302897, + "MSE/layer0": 627.595106302897, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.69, + "input_norm": 31.99773236592611, + "input_norm/layer0": 31.99773236592611, + "learning_rate": 0.00017894736842105264, + "loss": 1.9312, + "max_norm": 75.08959197998047, + "max_norm/layer0": 75.08959197998047, + "mean_norm": 64.3110408782959, + "mean_norm/layer0": 64.3110408782959, + "multicode_k": 1, + "output_norm": 14.375651826858522, + "output_norm/layer0": 14.375651826858522, + "step": 6600 + }, + { + "MSE": 627.2688003540036, + "MSE/layer0": 627.2688003540036, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.69, + "input_norm": 31.997724459966015, + "input_norm/layer0": 31.997724459966015, + "learning_rate": 0.0001763157894736842, + "loss": 1.9454, + "max_norm": 75.23365783691406, + "max_norm/layer0": 75.23365783691406, + "mean_norm": 64.42234230041504, + "mean_norm/layer0": 64.42234230041504, + "multicode_k": 1, + "output_norm": 14.385090745290121, + "output_norm/layer0": 14.385090745290121, + "step": 6650 + }, + { + "MSE": 626.5893623860678, + "MSE/layer0": 626.5893623860678, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.7, + "input_norm": 31.997720209757503, + "input_norm/layer0": 31.997720209757503, + "learning_rate": 0.0001736842105263158, + "loss": 1.9428, + "max_norm": 75.36791229248047, + "max_norm/layer0": 75.36791229248047, + "mean_norm": 64.5310287475586, + "mean_norm/layer0": 64.5310287475586, + "multicode_k": 1, + "output_norm": 14.414791498184208, + "output_norm/layer0": 14.414791498184208, + "step": 6700 + }, + { + "MSE": 626.1687516276043, + "MSE/layer0": 626.1687516276043, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.7, + "input_norm": 31.997717237472536, + "input_norm/layer0": 31.997717237472536, + "learning_rate": 0.00017105263157894739, + "loss": 1.9341, + "max_norm": 75.49561309814453, + "max_norm/layer0": 75.49561309814453, + "mean_norm": 64.63836669921875, + "mean_norm/layer0": 64.63836669921875, + "multicode_k": 1, + "output_norm": 14.436859647432962, + "output_norm/layer0": 14.436859647432962, + "step": 6750 + }, + { + "MSE": 625.7842074584966, + "MSE/layer0": 625.7842074584966, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.71, + "input_norm": 31.997723042170207, + "input_norm/layer0": 31.997723042170207, + "learning_rate": 0.00016842105263157895, + "loss": 1.9391, + "max_norm": 75.62852478027344, + "max_norm/layer0": 75.62852478027344, + "mean_norm": 64.74386024475098, + "mean_norm/layer0": 64.74386024475098, + "multicode_k": 1, + "output_norm": 14.45211536884308, + "output_norm/layer0": 14.45211536884308, + "step": 6800 + }, + { + "MSE": 625.3583324178057, + "MSE/layer0": 625.3583324178057, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.72, + "input_norm": 31.997710723876956, + "input_norm/layer0": 31.997710723876956, + "learning_rate": 0.00016578947368421052, + "loss": 1.9214, + "max_norm": 75.7518081665039, + "max_norm/layer0": 75.7518081665039, + "mean_norm": 64.84785079956055, + "mean_norm/layer0": 64.84785079956055, + "multicode_k": 1, + "output_norm": 14.472083713213603, + "output_norm/layer0": 14.472083713213603, + "step": 6850 + }, + { + "MSE": 625.0808269246418, + "MSE/layer0": 625.0808269246418, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.72, + "input_norm": 31.997701005935667, + "input_norm/layer0": 31.997701005935667, + "learning_rate": 0.0001631578947368421, + "loss": 1.9248, + "max_norm": 75.8736343383789, + "max_norm/layer0": 75.8736343383789, + "mean_norm": 64.94989013671875, + "mean_norm/layer0": 64.94989013671875, + "multicode_k": 1, + "output_norm": 14.49320138454437, + "output_norm/layer0": 14.49320138454437, + "step": 6900 + }, + { + "MSE": 624.4893544514975, + "MSE/layer0": 624.4893544514975, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.73, + "input_norm": 31.997702992757166, + "input_norm/layer0": 31.997702992757166, + "learning_rate": 0.0001605263157894737, + "loss": 1.9357, + "max_norm": 75.99244689941406, + "max_norm/layer0": 75.99244689941406, + "mean_norm": 65.05006790161133, + "mean_norm/layer0": 65.05006790161133, + "multicode_k": 1, + "output_norm": 14.515017460187277, + "output_norm/layer0": 14.515017460187277, + "step": 6950 + }, + { + "MSE": 623.983821309408, + "MSE/layer0": 623.983821309408, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.73, + "input_norm": 31.997692581812547, + "input_norm/layer0": 31.997692581812547, + "learning_rate": 0.00015789473684210527, + "loss": 1.9256, + "max_norm": 76.1169204711914, + "max_norm/layer0": 76.1169204711914, + "mean_norm": 65.14841270446777, + "mean_norm/layer0": 65.14841270446777, + "multicode_k": 1, + "output_norm": 14.531605450312297, + "output_norm/layer0": 14.531605450312297, + "step": 7000 + }, + { + "epoch": 0.73, + "eval_MSE/layer0": 623.2726008245854, + "eval_accuracy": 0.5373965313049694, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.99768957905041, + "eval_loss": 1.9302037954330444, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 14.553397603295936, + "eval_runtime": 73.3018, + "eval_samples_per_second": 63.068, + "eval_steps_per_second": 7.885, + "step": 7000 + }, + { + "MSE": 623.8173256429034, + "MSE/layer0": 623.8173256429034, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.74, + "input_norm": 31.997689800262457, + "input_norm/layer0": 31.997689800262457, + "learning_rate": 0.00015526315789473686, + "loss": 1.9215, + "max_norm": 76.22943115234375, + "max_norm/layer0": 76.22943115234375, + "mean_norm": 65.2452278137207, + "mean_norm/layer0": 65.2452278137207, + "multicode_k": 1, + "output_norm": 14.544135572115584, + "output_norm/layer0": 14.544135572115584, + "step": 7050 + }, + { + "MSE": 623.4564833577472, + "MSE/layer0": 623.4564833577472, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.74, + "input_norm": 31.997697146733607, + "input_norm/layer0": 31.997697146733607, + "learning_rate": 0.00015263157894736842, + "loss": 1.9291, + "max_norm": 76.35796356201172, + "max_norm/layer0": 76.35796356201172, + "mean_norm": 65.33997344970703, + "mean_norm/layer0": 65.33997344970703, + "multicode_k": 1, + "output_norm": 14.557166822751359, + "output_norm/layer0": 14.557166822751359, + "step": 7100 + }, + { + "MSE": 622.3157424926754, + "MSE/layer0": 622.3157424926754, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.75, + "input_norm": 31.997690575917574, + "input_norm/layer0": 31.997690575917574, + "learning_rate": 0.00015, + "loss": 1.9272, + "max_norm": 76.47930145263672, + "max_norm/layer0": 76.47930145263672, + "mean_norm": 65.4333724975586, + "mean_norm/layer0": 65.4333724975586, + "multicode_k": 1, + "output_norm": 14.59491890271505, + "output_norm/layer0": 14.59491890271505, + "step": 7150 + }, + { + "MSE": 622.1008169555663, + "MSE/layer0": 622.1008169555663, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.75, + "input_norm": 31.997691469192503, + "input_norm/layer0": 31.997691469192503, + "learning_rate": 0.00014736842105263158, + "loss": 1.9421, + "max_norm": 76.5845947265625, + "max_norm/layer0": 76.5845947265625, + "mean_norm": 65.52462577819824, + "mean_norm/layer0": 65.52462577819824, + "multicode_k": 1, + "output_norm": 14.608456416130064, + "output_norm/layer0": 14.608456416130064, + "step": 7200 + }, + { + "MSE": 621.7943653361006, + "MSE/layer0": 621.7943653361006, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.76, + "input_norm": 31.997678140004478, + "input_norm/layer0": 31.997678140004478, + "learning_rate": 0.00014473684210526317, + "loss": 1.9221, + "max_norm": 76.68899536132812, + "max_norm/layer0": 76.68899536132812, + "mean_norm": 65.61434745788574, + "mean_norm/layer0": 65.61434745788574, + "multicode_k": 1, + "output_norm": 14.622403078079222, + "output_norm/layer0": 14.622403078079222, + "step": 7250 + }, + { + "MSE": 621.7445918782552, + "MSE/layer0": 621.7445918782552, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.76, + "input_norm": 31.997679424285884, + "input_norm/layer0": 31.997679424285884, + "learning_rate": 0.00014210526315789474, + "loss": 1.9172, + "max_norm": 76.79942321777344, + "max_norm/layer0": 76.79942321777344, + "mean_norm": 65.70241737365723, + "mean_norm/layer0": 65.70241737365723, + "multicode_k": 1, + "output_norm": 14.632240413029983, + "output_norm/layer0": 14.632240413029983, + "step": 7300 + }, + { + "MSE": 621.0073055013017, + "MSE/layer0": 621.0073055013017, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.77, + "input_norm": 31.997667986551914, + "input_norm/layer0": 31.997667986551914, + "learning_rate": 0.0001394736842105263, + "loss": 1.9187, + "max_norm": 76.90473937988281, + "max_norm/layer0": 76.90473937988281, + "mean_norm": 65.78865623474121, + "mean_norm/layer0": 65.78865623474121, + "multicode_k": 1, + "output_norm": 14.659644064903254, + "output_norm/layer0": 14.659644064903254, + "step": 7350 + }, + { + "MSE": 620.5166587320964, + "MSE/layer0": 620.5166587320964, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.77, + "input_norm": 31.99766827583312, + "input_norm/layer0": 31.99766827583312, + "learning_rate": 0.00013684210526315792, + "loss": 1.9236, + "max_norm": 77.00653839111328, + "max_norm/layer0": 77.00653839111328, + "mean_norm": 65.87344741821289, + "mean_norm/layer0": 65.87344741821289, + "multicode_k": 1, + "output_norm": 14.683248674074807, + "output_norm/layer0": 14.683248674074807, + "step": 7400 + }, + { + "MSE": 620.4730934651691, + "MSE/layer0": 620.4730934651691, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.78, + "input_norm": 31.99766536712645, + "input_norm/layer0": 31.99766536712645, + "learning_rate": 0.00013421052631578948, + "loss": 1.9181, + "max_norm": 77.11151123046875, + "max_norm/layer0": 77.11151123046875, + "mean_norm": 65.95642852783203, + "mean_norm/layer0": 65.95642852783203, + "multicode_k": 1, + "output_norm": 14.682427426973977, + "output_norm/layer0": 14.682427426973977, + "step": 7450 + }, + { + "MSE": 619.8806704711913, + "MSE/layer0": 619.8806704711913, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.78, + "input_norm": 31.997652931213374, + "input_norm/layer0": 31.997652931213374, + "learning_rate": 0.00013157894736842105, + "loss": 1.9204, + "max_norm": 77.21614837646484, + "max_norm/layer0": 77.21614837646484, + "mean_norm": 66.03750610351562, + "mean_norm/layer0": 66.03750610351562, + "multicode_k": 1, + "output_norm": 14.709125100771587, + "output_norm/layer0": 14.709125100771587, + "step": 7500 + }, + { + "epoch": 0.78, + "eval_MSE/layer0": 619.4572802491444, + "eval_accuracy": 0.538146743438657, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997657016941467, + "eval_loss": 1.9224542379379272, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 14.72584700899003, + "eval_runtime": 73.2809, + "eval_samples_per_second": 63.086, + "eval_steps_per_second": 7.887, + "step": 7500 + }, + { + "MSE": 619.6498880004883, + "MSE/layer0": 619.6498880004883, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.79, + "input_norm": 31.997653865814208, + "input_norm/layer0": 31.997653865814208, + "learning_rate": 0.00012894736842105264, + "loss": 1.9109, + "max_norm": 77.3195571899414, + "max_norm/layer0": 77.3195571899414, + "mean_norm": 66.11709403991699, + "mean_norm/layer0": 66.11709403991699, + "multicode_k": 1, + "output_norm": 14.724224853515622, + "output_norm/layer0": 14.724224853515622, + "step": 7550 + }, + { + "MSE": 619.544646809896, + "MSE/layer0": 619.544646809896, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.79, + "input_norm": 31.997655792236333, + "input_norm/layer0": 31.997655792236333, + "learning_rate": 0.0001263157894736842, + "loss": 1.9247, + "max_norm": 77.41654205322266, + "max_norm/layer0": 77.41654205322266, + "mean_norm": 66.19502639770508, + "mean_norm/layer0": 66.19502639770508, + "multicode_k": 1, + "output_norm": 14.729852019945778, + "output_norm/layer0": 14.729852019945778, + "step": 7600 + }, + { + "MSE": 619.1442233276366, + "MSE/layer0": 619.1442233276366, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.8, + "input_norm": 31.99764471054077, + "input_norm/layer0": 31.99764471054077, + "learning_rate": 0.0001236842105263158, + "loss": 1.9237, + "max_norm": 77.5074234008789, + "max_norm/layer0": 77.5074234008789, + "mean_norm": 66.27114677429199, + "mean_norm/layer0": 66.27114677429199, + "multicode_k": 1, + "output_norm": 14.745990212758379, + "output_norm/layer0": 14.745990212758379, + "step": 7650 + }, + { + "MSE": 618.6404962158206, + "MSE/layer0": 618.6404962158206, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.8, + "input_norm": 31.997635892232267, + "input_norm/layer0": 31.997635892232267, + "learning_rate": 0.00012105263157894738, + "loss": 1.913, + "max_norm": 77.602294921875, + "max_norm/layer0": 77.602294921875, + "mean_norm": 66.34577751159668, + "mean_norm/layer0": 66.34577751159668, + "multicode_k": 1, + "output_norm": 14.766639779408772, + "output_norm/layer0": 14.766639779408772, + "step": 7700 + }, + { + "MSE": 618.2833578491213, + "MSE/layer0": 618.2833578491213, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.81, + "input_norm": 31.997631740570075, + "input_norm/layer0": 31.997631740570075, + "learning_rate": 0.00011842105263157894, + "loss": 1.9214, + "max_norm": 77.6917724609375, + "max_norm/layer0": 77.6917724609375, + "mean_norm": 66.41888046264648, + "mean_norm/layer0": 66.41888046264648, + "multicode_k": 1, + "output_norm": 14.779039435386654, + "output_norm/layer0": 14.779039435386654, + "step": 7750 + }, + { + "MSE": 618.2477112833653, + "MSE/layer0": 618.2477112833653, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.81, + "input_norm": 31.997634382247924, + "input_norm/layer0": 31.997634382247924, + "learning_rate": 0.00011578947368421053, + "loss": 1.9127, + "max_norm": 77.77839660644531, + "max_norm/layer0": 77.77839660644531, + "mean_norm": 66.49017333984375, + "mean_norm/layer0": 66.49017333984375, + "multicode_k": 1, + "output_norm": 14.782011265754704, + "output_norm/layer0": 14.782011265754704, + "step": 7800 + }, + { + "MSE": 617.7417582194005, + "MSE/layer0": 617.7417582194005, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.82, + "input_norm": 31.997628266016648, + "input_norm/layer0": 31.997628266016648, + "learning_rate": 0.00011315789473684211, + "loss": 1.9084, + "max_norm": 77.86212158203125, + "max_norm/layer0": 77.86212158203125, + "mean_norm": 66.55990791320801, + "mean_norm/layer0": 66.55990791320801, + "multicode_k": 1, + "output_norm": 14.801776518821718, + "output_norm/layer0": 14.801776518821718, + "step": 7850 + }, + { + "MSE": 617.339886271159, + "MSE/layer0": 617.339886271159, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.82, + "input_norm": 31.99762384732564, + "input_norm/layer0": 31.99762384732564, + "learning_rate": 0.00011052631578947368, + "loss": 1.9115, + "max_norm": 77.94374084472656, + "max_norm/layer0": 77.94374084472656, + "mean_norm": 66.62779235839844, + "mean_norm/layer0": 66.62779235839844, + "multicode_k": 1, + "output_norm": 14.823196705182394, + "output_norm/layer0": 14.823196705182394, + "step": 7900 + }, + { + "MSE": 617.3184334309897, + "MSE/layer0": 617.3184334309897, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.83, + "input_norm": 31.9976179567973, + "input_norm/layer0": 31.9976179567973, + "learning_rate": 0.00010789473684210527, + "loss": 1.9136, + "max_norm": 78.02580261230469, + "max_norm/layer0": 78.02580261230469, + "mean_norm": 66.69412803649902, + "mean_norm/layer0": 66.69412803649902, + "multicode_k": 1, + "output_norm": 14.828250519434608, + "output_norm/layer0": 14.828250519434608, + "step": 7950 + }, + { + "MSE": 616.9322255452475, + "MSE/layer0": 616.9322255452475, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.84, + "input_norm": 31.997613105773937, + "input_norm/layer0": 31.997613105773937, + "learning_rate": 0.00010526315789473683, + "loss": 1.907, + "max_norm": 78.10686492919922, + "max_norm/layer0": 78.10686492919922, + "mean_norm": 66.7584114074707, + "mean_norm/layer0": 66.7584114074707, + "multicode_k": 1, + "output_norm": 14.839720834096273, + "output_norm/layer0": 14.839720834096273, + "step": 8000 + }, + { + "epoch": 0.84, + "eval_MSE/layer0": 616.4379357749087, + "eval_accuracy": 0.5393073732024142, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.99761064584294, + "eval_loss": 1.9150168895721436, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 14.862492301828695, + "eval_runtime": 73.6278, + "eval_samples_per_second": 62.789, + "eval_steps_per_second": 7.85, + "step": 8000 + }, + { + "MSE": 616.6813212076825, + "MSE/layer0": 616.6813212076825, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.84, + "input_norm": 31.997603750228897, + "input_norm/layer0": 31.997603750228897, + "learning_rate": 0.00010263157894736843, + "loss": 1.8975, + "max_norm": 78.18397521972656, + "max_norm/layer0": 78.18397521972656, + "mean_norm": 66.82158279418945, + "mean_norm/layer0": 66.82158279418945, + "multicode_k": 1, + "output_norm": 14.848202861150106, + "output_norm/layer0": 14.848202861150106, + "step": 8050 + }, + { + "MSE": 616.5551970418294, + "MSE/layer0": 616.5551970418294, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.85, + "input_norm": 31.99760689099629, + "input_norm/layer0": 31.99760689099629, + "learning_rate": 0.0001, + "loss": 1.916, + "max_norm": 78.26499938964844, + "max_norm/layer0": 78.26499938964844, + "mean_norm": 66.88335037231445, + "mean_norm/layer0": 66.88335037231445, + "multicode_k": 1, + "output_norm": 14.8604402812322, + "output_norm/layer0": 14.8604402812322, + "step": 8100 + }, + { + "MSE": 616.288039347331, + "MSE/layer0": 616.288039347331, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.85, + "input_norm": 31.997600466410333, + "input_norm/layer0": 31.997600466410333, + "learning_rate": 9.736842105263158e-05, + "loss": 1.902, + "max_norm": 78.33844757080078, + "max_norm/layer0": 78.33844757080078, + "mean_norm": 66.94340133666992, + "mean_norm/layer0": 66.94340133666992, + "multicode_k": 1, + "output_norm": 14.872187639872232, + "output_norm/layer0": 14.872187639872232, + "step": 8150 + }, + { + "MSE": 615.8982196044924, + "MSE/layer0": 615.8982196044924, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.86, + "input_norm": 31.997594718933108, + "input_norm/layer0": 31.997594718933108, + "learning_rate": 9.473684210526316e-05, + "loss": 1.9142, + "max_norm": 78.40998077392578, + "max_norm/layer0": 78.40998077392578, + "mean_norm": 67.00171661376953, + "mean_norm/layer0": 67.00171661376953, + "multicode_k": 1, + "output_norm": 14.884622203509018, + "output_norm/layer0": 14.884622203509018, + "step": 8200 + }, + { + "MSE": 615.649053141276, + "MSE/layer0": 615.649053141276, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.86, + "input_norm": 31.997591203053794, + "input_norm/layer0": 31.997591203053794, + "learning_rate": 9.210526315789474e-05, + "loss": 1.9103, + "max_norm": 78.47700500488281, + "max_norm/layer0": 78.47700500488281, + "mean_norm": 67.05831527709961, + "mean_norm/layer0": 67.05831527709961, + "multicode_k": 1, + "output_norm": 14.896942078272502, + "output_norm/layer0": 14.896942078272502, + "step": 8250 + }, + { + "MSE": 615.4050069173176, + "MSE/layer0": 615.4050069173176, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.87, + "input_norm": 31.99757507324218, + "input_norm/layer0": 31.99757507324218, + "learning_rate": 8.947368421052632e-05, + "loss": 1.8999, + "max_norm": 78.54086303710938, + "max_norm/layer0": 78.54086303710938, + "mean_norm": 67.11351013183594, + "mean_norm/layer0": 67.11351013183594, + "multicode_k": 1, + "output_norm": 14.907591681480406, + "output_norm/layer0": 14.907591681480406, + "step": 8300 + }, + { + "MSE": 615.0221789550782, + "MSE/layer0": 615.0221789550782, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.87, + "input_norm": 31.997587076822917, + "input_norm/layer0": 31.997587076822917, + "learning_rate": 8.68421052631579e-05, + "loss": 1.9122, + "max_norm": 78.60425567626953, + "max_norm/layer0": 78.60425567626953, + "mean_norm": 67.1669692993164, + "mean_norm/layer0": 67.1669692993164, + "multicode_k": 1, + "output_norm": 14.918850135803218, + "output_norm/layer0": 14.918850135803218, + "step": 8350 + }, + { + "MSE": 614.7660255940759, + "MSE/layer0": 614.7660255940759, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.88, + "input_norm": 31.99758012771608, + "input_norm/layer0": 31.99758012771608, + "learning_rate": 8.421052631578948e-05, + "loss": 1.9074, + "max_norm": 78.66250610351562, + "max_norm/layer0": 78.66250610351562, + "mean_norm": 67.21884536743164, + "mean_norm/layer0": 67.21884536743164, + "multicode_k": 1, + "output_norm": 14.930259111722311, + "output_norm/layer0": 14.930259111722311, + "step": 8400 + }, + { + "MSE": 614.4904387410484, + "MSE/layer0": 614.4904387410484, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.88, + "input_norm": 31.99757884025574, + "input_norm/layer0": 31.99757884025574, + "learning_rate": 8.157894736842105e-05, + "loss": 1.9151, + "max_norm": 78.7247314453125, + "max_norm/layer0": 78.7247314453125, + "mean_norm": 67.26914596557617, + "mean_norm/layer0": 67.26914596557617, + "multicode_k": 1, + "output_norm": 14.941800510088594, + "output_norm/layer0": 14.941800510088594, + "step": 8450 + }, + { + "MSE": 614.3984759521479, + "MSE/layer0": 614.3984759521479, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.89, + "input_norm": 31.997565978368122, + "input_norm/layer0": 31.997565978368122, + "learning_rate": 7.894736842105263e-05, + "loss": 1.8931, + "max_norm": 78.78428649902344, + "max_norm/layer0": 78.78428649902344, + "mean_norm": 67.31785583496094, + "mean_norm/layer0": 67.31785583496094, + "multicode_k": 1, + "output_norm": 14.948297271728517, + "output_norm/layer0": 14.948297271728517, + "step": 8500 + }, + { + "epoch": 0.89, + "eval_MSE/layer0": 613.78736410403, + "eval_accuracy": 0.5408171011151899, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997572115378908, + "eval_loss": 1.9076036214828491, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 14.968526063531659, + "eval_runtime": 73.9448, + "eval_samples_per_second": 62.52, + "eval_steps_per_second": 7.817, + "step": 8500 + }, + { + "MSE": 614.2003710937502, + "MSE/layer0": 614.2003710937502, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.89, + "input_norm": 31.997571328481037, + "input_norm/layer0": 31.997571328481037, + "learning_rate": 7.631578947368421e-05, + "loss": 1.9006, + "max_norm": 78.83836364746094, + "max_norm/layer0": 78.83836364746094, + "mean_norm": 67.36493301391602, + "mean_norm/layer0": 67.36493301391602, + "multicode_k": 1, + "output_norm": 14.955024781227117, + "output_norm/layer0": 14.955024781227117, + "step": 8550 + }, + { + "MSE": 613.8168900553383, + "MSE/layer0": 613.8168900553383, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.9, + "input_norm": 31.997557487487796, + "input_norm/layer0": 31.997557487487796, + "learning_rate": 7.368421052631579e-05, + "loss": 1.9045, + "max_norm": 78.8912582397461, + "max_norm/layer0": 78.8912582397461, + "mean_norm": 67.41046524047852, + "mean_norm/layer0": 67.41046524047852, + "multicode_k": 1, + "output_norm": 14.968488362630207, + "output_norm/layer0": 14.968488362630207, + "step": 8600 + }, + { + "MSE": 613.6968625895179, + "MSE/layer0": 613.6968625895179, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.9, + "input_norm": 31.99754828453064, + "input_norm/layer0": 31.99754828453064, + "learning_rate": 7.105263157894737e-05, + "loss": 1.9009, + "max_norm": 78.942626953125, + "max_norm/layer0": 78.942626953125, + "mean_norm": 67.45438766479492, + "mean_norm/layer0": 67.45438766479492, + "multicode_k": 1, + "output_norm": 14.979475774765014, + "output_norm/layer0": 14.979475774765014, + "step": 8650 + }, + { + "MSE": 613.3956824747725, + "MSE/layer0": 613.3956824747725, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.91, + "input_norm": 31.997546965281174, + "input_norm/layer0": 31.997546965281174, + "learning_rate": 6.842105263157896e-05, + "loss": 1.9, + "max_norm": 78.99479675292969, + "max_norm/layer0": 78.99479675292969, + "mean_norm": 67.49666595458984, + "mean_norm/layer0": 67.49666595458984, + "multicode_k": 1, + "output_norm": 14.988234910964966, + "output_norm/layer0": 14.988234910964966, + "step": 8700 + }, + { + "MSE": 613.2128627522789, + "MSE/layer0": 613.2128627522789, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.91, + "input_norm": 31.997544927597048, + "input_norm/layer0": 31.997544927597048, + "learning_rate": 6.578947368421052e-05, + "loss": 1.9059, + "max_norm": 79.04541015625, + "max_norm/layer0": 79.04541015625, + "mean_norm": 67.53742218017578, + "mean_norm/layer0": 67.53742218017578, + "multicode_k": 1, + "output_norm": 14.991036421457924, + "output_norm/layer0": 14.991036421457924, + "step": 8750 + }, + { + "MSE": 612.9370720418297, + "MSE/layer0": 612.9370720418297, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.92, + "input_norm": 31.99753908475239, + "input_norm/layer0": 31.99753908475239, + "learning_rate": 6.31578947368421e-05, + "loss": 1.9023, + "max_norm": 79.09040069580078, + "max_norm/layer0": 79.09040069580078, + "mean_norm": 67.57658767700195, + "mean_norm/layer0": 67.57658767700195, + "multicode_k": 1, + "output_norm": 15.003661061922706, + "output_norm/layer0": 15.003661061922706, + "step": 8800 + }, + { + "MSE": 613.0978963216148, + "MSE/layer0": 613.0978963216148, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.92, + "input_norm": 31.997535756429023, + "input_norm/layer0": 31.997535756429023, + "learning_rate": 6.052631578947369e-05, + "loss": 1.9004, + "max_norm": 79.13478088378906, + "max_norm/layer0": 79.13478088378906, + "mean_norm": 67.61412811279297, + "mean_norm/layer0": 67.61412811279297, + "multicode_k": 1, + "output_norm": 14.999437109629307, + "output_norm/layer0": 14.999437109629307, + "step": 8850 + }, + { + "MSE": 612.746408691406, + "MSE/layer0": 612.746408691406, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.93, + "input_norm": 31.997531512578334, + "input_norm/layer0": 31.997531512578334, + "learning_rate": 5.789473684210527e-05, + "loss": 1.8947, + "max_norm": 79.17863464355469, + "max_norm/layer0": 79.17863464355469, + "mean_norm": 67.65010452270508, + "mean_norm/layer0": 67.65010452270508, + "multicode_k": 1, + "output_norm": 15.013854147593182, + "output_norm/layer0": 15.013854147593182, + "step": 8900 + }, + { + "MSE": 612.5075473022462, + "MSE/layer0": 612.5075473022462, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.93, + "input_norm": 31.997522573471066, + "input_norm/layer0": 31.997522573471066, + "learning_rate": 5.526315789473684e-05, + "loss": 1.888, + "max_norm": 79.2198257446289, + "max_norm/layer0": 79.2198257446289, + "mean_norm": 67.6845588684082, + "mean_norm/layer0": 67.6845588684082, + "multicode_k": 1, + "output_norm": 15.024005990028382, + "output_norm/layer0": 15.024005990028382, + "step": 8950 + }, + { + "MSE": 612.4464337158204, + "MSE/layer0": 612.4464337158204, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.94, + "input_norm": 31.99751985549927, + "input_norm/layer0": 31.99751985549927, + "learning_rate": 5.263157894736842e-05, + "loss": 1.9021, + "max_norm": 79.25985717773438, + "max_norm/layer0": 79.25985717773438, + "mean_norm": 67.71733856201172, + "mean_norm/layer0": 67.71733856201172, + "multicode_k": 1, + "output_norm": 15.025202210744226, + "output_norm/layer0": 15.025202210744226, + "step": 9000 + }, + { + "epoch": 0.94, + "eval_MSE/layer0": 612.012579843437, + "eval_accuracy": 0.5416772654217966, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.99751990196794, + "eval_loss": 1.90205979347229, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 15.037853428586699, + "eval_runtime": 73.2981, + "eval_samples_per_second": 63.071, + "eval_steps_per_second": 7.886, + "step": 9000 + }, + { + "MSE": 612.1616466267901, + "MSE/layer0": 612.1616466267901, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.94, + "input_norm": 31.997515144348153, + "input_norm/layer0": 31.997515144348153, + "learning_rate": 5e-05, + "loss": 1.8979, + "max_norm": 79.2950668334961, + "max_norm/layer0": 79.2950668334961, + "mean_norm": 67.74863052368164, + "mean_norm/layer0": 67.74863052368164, + "multicode_k": 1, + "output_norm": 15.036479252179465, + "output_norm/layer0": 15.036479252179465, + "step": 9050 + }, + { + "MSE": 611.8442991129552, + "MSE/layer0": 611.8442991129552, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.95, + "input_norm": 31.99751302719116, + "input_norm/layer0": 31.99751302719116, + "learning_rate": 4.736842105263158e-05, + "loss": 1.8978, + "max_norm": 79.32892608642578, + "max_norm/layer0": 79.32892608642578, + "mean_norm": 67.77827835083008, + "mean_norm/layer0": 67.77827835083008, + "multicode_k": 1, + "output_norm": 15.046743833223978, + "output_norm/layer0": 15.046743833223978, + "step": 9100 + }, + { + "MSE": 611.9183032226562, + "MSE/layer0": 611.9183032226562, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.96, + "input_norm": 31.99751057942708, + "input_norm/layer0": 31.99751057942708, + "learning_rate": 4.473684210526316e-05, + "loss": 1.8971, + "max_norm": 79.36182403564453, + "max_norm/layer0": 79.36182403564453, + "mean_norm": 67.80632781982422, + "mean_norm/layer0": 67.80632781982422, + "multicode_k": 1, + "output_norm": 15.041637244224557, + "output_norm/layer0": 15.041637244224557, + "step": 9150 + }, + { + "MSE": 611.5441438802083, + "MSE/layer0": 611.5441438802083, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.96, + "input_norm": 31.99750095685323, + "input_norm/layer0": 31.99750095685323, + "learning_rate": 4.210526315789474e-05, + "loss": 1.8874, + "max_norm": 79.39281463623047, + "max_norm/layer0": 79.39281463623047, + "mean_norm": 67.83284759521484, + "mean_norm/layer0": 67.83284759521484, + "multicode_k": 1, + "output_norm": 15.055660729408274, + "output_norm/layer0": 15.055660729408274, + "step": 9200 + }, + { + "MSE": 611.5922235107425, + "MSE/layer0": 611.5922235107425, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.97, + "input_norm": 31.99750220934551, + "input_norm/layer0": 31.99750220934551, + "learning_rate": 3.9473684210526316e-05, + "loss": 1.8958, + "max_norm": 79.42273712158203, + "max_norm/layer0": 79.42273712158203, + "mean_norm": 67.85774230957031, + "mean_norm/layer0": 67.85774230957031, + "multicode_k": 1, + "output_norm": 15.055747102101643, + "output_norm/layer0": 15.055747102101643, + "step": 9250 + }, + { + "MSE": 611.6544079589839, + "MSE/layer0": 611.6544079589839, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.97, + "input_norm": 31.997499033610026, + "input_norm/layer0": 31.997499033610026, + "learning_rate": 3.6842105263157895e-05, + "loss": 1.8915, + "max_norm": 79.44976806640625, + "max_norm/layer0": 79.44976806640625, + "mean_norm": 67.88099670410156, + "mean_norm/layer0": 67.88099670410156, + "multicode_k": 1, + "output_norm": 15.057963668505355, + "output_norm/layer0": 15.057963668505355, + "step": 9300 + }, + { + "MSE": 611.321997172038, + "MSE/layer0": 611.321997172038, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.98, + "input_norm": 31.997498153050746, + "input_norm/layer0": 31.997498153050746, + "learning_rate": 3.421052631578948e-05, + "loss": 1.8893, + "max_norm": 79.47447967529297, + "max_norm/layer0": 79.47447967529297, + "mean_norm": 67.90266799926758, + "mean_norm/layer0": 67.90266799926758, + "multicode_k": 1, + "output_norm": 15.067080327669775, + "output_norm/layer0": 15.067080327669775, + "step": 9350 + }, + { + "MSE": 611.4500786336266, + "MSE/layer0": 611.4500786336266, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.98, + "input_norm": 31.997495447794595, + "input_norm/layer0": 31.997495447794595, + "learning_rate": 3.157894736842105e-05, + "loss": 1.894, + "max_norm": 79.49812316894531, + "max_norm/layer0": 79.49812316894531, + "mean_norm": 67.92279815673828, + "mean_norm/layer0": 67.92279815673828, + "multicode_k": 1, + "output_norm": 15.062444001833596, + "output_norm/layer0": 15.062444001833596, + "step": 9400 + }, + { + "MSE": 611.1107730102539, + "MSE/layer0": 611.1107730102539, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.99, + "input_norm": 31.997485151290896, + "input_norm/layer0": 31.997485151290896, + "learning_rate": 2.8947368421052634e-05, + "loss": 1.8849, + "max_norm": 79.52143096923828, + "max_norm/layer0": 79.52143096923828, + "mean_norm": 67.94132995605469, + "mean_norm/layer0": 67.94132995605469, + "multicode_k": 1, + "output_norm": 15.076274760564168, + "output_norm/layer0": 15.076274760564168, + "step": 9450 + }, + { + "MSE": 611.3009430948896, + "MSE/layer0": 611.3009430948896, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 0.99, + "input_norm": 31.99749323209126, + "input_norm/layer0": 31.99749323209126, + "learning_rate": 2.631578947368421e-05, + "loss": 1.8967, + "max_norm": 79.54227447509766, + "max_norm/layer0": 79.54227447509766, + "mean_norm": 67.958251953125, + "mean_norm/layer0": 67.958251953125, + "multicode_k": 1, + "output_norm": 15.06888332684835, + "output_norm/layer0": 15.06888332684835, + "step": 9500 + }, + { + "epoch": 0.99, + "eval_MSE/layer0": 610.6120883183328, + "eval_accuracy": 0.5425511737500183, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.99749166347134, + "eval_loss": 1.8969556093215942, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 15.09320597480496, + "eval_runtime": 73.3984, + "eval_samples_per_second": 62.985, + "eval_steps_per_second": 7.875, + "step": 9500 + }, + { + "MSE": 610.9202908325196, + "MSE/layer0": 610.9202908325196, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.0, + "input_norm": 31.99748815218606, + "input_norm/layer0": 31.99748815218606, + "learning_rate": 2.368421052631579e-05, + "loss": 1.8917, + "max_norm": 79.56092834472656, + "max_norm/layer0": 79.56092834472656, + "mean_norm": 67.97361755371094, + "mean_norm/layer0": 67.97361755371094, + "multicode_k": 1, + "output_norm": 15.084220841725665, + "output_norm/layer0": 15.084220841725665, + "step": 9550 + }, + { + "MSE": 610.9847631835939, + "MSE/layer0": 610.9847631835939, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.0, + "input_norm": 31.997486731211332, + "input_norm/layer0": 31.997486731211332, + "learning_rate": 2.105263157894737e-05, + "loss": 1.8839, + "max_norm": 79.57735443115234, + "max_norm/layer0": 79.57735443115234, + "mean_norm": 67.98743438720703, + "mean_norm/layer0": 67.98743438720703, + "multicode_k": 1, + "output_norm": 15.082832886377968, + "output_norm/layer0": 15.082832886377968, + "step": 9600 + }, + { + "MSE": 611.2879392496747, + "MSE/layer0": 611.2879392496747, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.01, + "input_norm": 31.997482639948544, + "input_norm/layer0": 31.997482639948544, + "learning_rate": 1.8421052631578947e-05, + "loss": 1.8851, + "max_norm": 79.59221649169922, + "max_norm/layer0": 79.59221649169922, + "mean_norm": 67.99962997436523, + "mean_norm/layer0": 67.99962997436523, + "multicode_k": 1, + "output_norm": 15.075549699465444, + "output_norm/layer0": 15.075549699465444, + "step": 9650 + }, + { + "MSE": 611.3861442057291, + "MSE/layer0": 611.3861442057291, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.01, + "input_norm": 31.997482592264817, + "input_norm/layer0": 31.997482592264817, + "learning_rate": 1.5789473684210526e-05, + "loss": 1.8774, + "max_norm": 79.60480499267578, + "max_norm/layer0": 79.60480499267578, + "mean_norm": 68.01019668579102, + "mean_norm/layer0": 68.01019668579102, + "multicode_k": 1, + "output_norm": 15.07396024545034, + "output_norm/layer0": 15.07396024545034, + "step": 9700 + }, + { + "MSE": 611.4255168660482, + "MSE/layer0": 611.4255168660482, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.02, + "input_norm": 31.997479289372762, + "input_norm/layer0": 31.997479289372762, + "learning_rate": 1.3157894736842104e-05, + "loss": 1.8768, + "max_norm": 79.6154556274414, + "max_norm/layer0": 79.6154556274414, + "mean_norm": 68.01911926269531, + "mean_norm/layer0": 68.01911926269531, + "multicode_k": 1, + "output_norm": 15.07339178085327, + "output_norm/layer0": 15.07339178085327, + "step": 9750 + }, + { + "MSE": 611.6131436157225, + "MSE/layer0": 611.6131436157225, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.02, + "input_norm": 31.99748600323995, + "input_norm/layer0": 31.99748600323995, + "learning_rate": 1.0526315789473684e-05, + "loss": 1.8905, + "max_norm": 79.62410736083984, + "max_norm/layer0": 79.62410736083984, + "mean_norm": 68.02641677856445, + "mean_norm/layer0": 68.02641677856445, + "multicode_k": 1, + "output_norm": 15.068124500910447, + "output_norm/layer0": 15.068124500910447, + "step": 9800 + }, + { + "MSE": 611.5507637532555, + "MSE/layer0": 611.5507637532555, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.03, + "input_norm": 31.997482582728068, + "input_norm/layer0": 31.997482582728068, + "learning_rate": 7.894736842105263e-06, + "loss": 1.8798, + "max_norm": 79.63082122802734, + "max_norm/layer0": 79.63082122802734, + "mean_norm": 68.03211212158203, + "mean_norm/layer0": 68.03211212158203, + "multicode_k": 1, + "output_norm": 15.072520554860436, + "output_norm/layer0": 15.072520554860436, + "step": 9850 + }, + { + "MSE": 611.7908610026044, + "MSE/layer0": 611.7908610026044, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.03, + "input_norm": 31.99747860272725, + "input_norm/layer0": 31.99747860272725, + "learning_rate": 5.263157894736842e-06, + "loss": 1.8807, + "max_norm": 79.63563537597656, + "max_norm/layer0": 79.63563537597656, + "mean_norm": 68.03619003295898, + "mean_norm/layer0": 68.03619003295898, + "multicode_k": 1, + "output_norm": 15.06489105542501, + "output_norm/layer0": 15.06489105542501, + "step": 9900 + }, + { + "MSE": 611.5220219930011, + "MSE/layer0": 611.5220219930011, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.04, + "input_norm": 31.997478488286326, + "input_norm/layer0": 31.997478488286326, + "learning_rate": 2.631578947368421e-06, + "loss": 1.8795, + "max_norm": 79.63849639892578, + "max_norm/layer0": 79.63849639892578, + "mean_norm": 68.03863906860352, + "mean_norm/layer0": 68.03863906860352, + "multicode_k": 1, + "output_norm": 15.07397619565328, + "output_norm/layer0": 15.07397619565328, + "step": 9950 + }, + { + "MSE": 611.5742947387696, + "MSE/layer0": 611.5742947387696, + "dead_code_fraction": 0.0, + "dead_code_fraction/layer0": 0.0, + "epoch": 1.04, + "input_norm": 31.997486855189003, + "input_norm/layer0": 31.997486855189003, + "learning_rate": 0.0, + "loss": 1.8942, + "max_norm": 79.63946533203125, + "max_norm/layer0": 79.63946533203125, + "mean_norm": 68.03947448730469, + "mean_norm/layer0": 68.03947448730469, + "multicode_k": 1, + "output_norm": 15.069696005185442, + "output_norm/layer0": 15.069696005185442, + "step": 10000 + }, + { + "epoch": 1.04, + "eval_MSE/layer0": 611.1571513346564, + "eval_accuracy": 0.5429091526514649, + "eval_dead_code_fraction/layer0": 0.0, + "eval_input_norm/layer0": 31.997479090978388, + "eval_loss": 1.89570152759552, + "eval_multicode_k": 1, + "eval_output_norm/layer0": 15.087154228553715, + "eval_runtime": 73.2125, + "eval_samples_per_second": 63.145, + "eval_steps_per_second": 7.895, + "step": 10000 + }, + { + "MSE": 0.0, + "MSE/layer0": 0.0, + "dead_code_fraction": 1.0, + "dead_code_fraction/layer0": 1.0, + "epoch": 1.04, + "input_norm": 0.0, + "input_norm/layer0": 0.0, + "max_norm": 79.63946533203125, + "max_norm/layer0": 79.63946533203125, + "mean_norm": 68.03947448730469, + "mean_norm/layer0": 68.03947448730469, + "multicode_k": 1, + "output_norm": 0.0, + "output_norm/layer0": 0.0, + "step": 10000, + "total_flos": 3.715683581952e+16, + "train_loss": 2.0762174885749816, + "train_runtime": 12054.7701, + "train_samples_per_second": 39.818, + "train_steps_per_second": 0.83 + } + ], + "logging_steps": 50, + "max_steps": 10000, + "num_train_epochs": 2, + "save_steps": 500, + "total_flos": 3.715683581952e+16, + "trial_name": null, + "trial_params": null +}