| { | |
| "best_metric": 0.46308374404907227, | |
| "best_model_checkpoint": "wav2vec2-xls-r-300m-fula/checkpoint-4000", | |
| "epoch": 5.154639175257732, | |
| "eval_steps": 2000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01288659793814433, | |
| "grad_norm": 10.90772533416748, | |
| "learning_rate": 3.4499999999999996e-06, | |
| "loss": 19.1295, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02577319587628866, | |
| "grad_norm": 12.727262496948242, | |
| "learning_rate": 7.2e-06, | |
| "loss": 16.6655, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03865979381443299, | |
| "grad_norm": 17.578950881958008, | |
| "learning_rate": 1.0949999999999998e-05, | |
| "loss": 13.462, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05154639175257732, | |
| "grad_norm": 14.574068069458008, | |
| "learning_rate": 1.47e-05, | |
| "loss": 7.4496, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06443298969072164, | |
| "grad_norm": 11.906006813049316, | |
| "learning_rate": 1.8449999999999998e-05, | |
| "loss": 6.1341, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07731958762886598, | |
| "grad_norm": 10.743711471557617, | |
| "learning_rate": 2.2199999999999998e-05, | |
| "loss": 5.0479, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09020618556701031, | |
| "grad_norm": 9.327588081359863, | |
| "learning_rate": 2.5949999999999997e-05, | |
| "loss": 4.86, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.10309278350515463, | |
| "grad_norm": 8.118748664855957, | |
| "learning_rate": 2.97e-05, | |
| "loss": 4.3188, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11597938144329897, | |
| "grad_norm": 4.968245983123779, | |
| "learning_rate": 3.345e-05, | |
| "loss": 4.142, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.12886597938144329, | |
| "grad_norm": 3.5491788387298584, | |
| "learning_rate": 3.7199999999999996e-05, | |
| "loss": 3.7329, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.14175257731958762, | |
| "grad_norm": 2.228337049484253, | |
| "learning_rate": 4.095e-05, | |
| "loss": 3.5432, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.15463917525773196, | |
| "grad_norm": 1.3472025394439697, | |
| "learning_rate": 4.4699999999999996e-05, | |
| "loss": 3.3229, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16752577319587628, | |
| "grad_norm": 0.8919354677200317, | |
| "learning_rate": 4.845e-05, | |
| "loss": 3.1852, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.18041237113402062, | |
| "grad_norm": 0.9615734815597534, | |
| "learning_rate": 5.2199999999999995e-05, | |
| "loss": 3.0339, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.19329896907216496, | |
| "grad_norm": 1.0601171255111694, | |
| "learning_rate": 5.595e-05, | |
| "loss": 2.9307, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.20618556701030927, | |
| "grad_norm": 1.2133561372756958, | |
| "learning_rate": 5.97e-05, | |
| "loss": 2.7395, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2190721649484536, | |
| "grad_norm": 0.922024130821228, | |
| "learning_rate": 6.345e-05, | |
| "loss": 2.5765, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.23195876288659795, | |
| "grad_norm": 1.0528242588043213, | |
| "learning_rate": 6.72e-05, | |
| "loss": 2.2256, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.24484536082474226, | |
| "grad_norm": 1.39167058467865, | |
| "learning_rate": 7.094999999999999e-05, | |
| "loss": 1.7335, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.25773195876288657, | |
| "grad_norm": 1.182673692703247, | |
| "learning_rate": 7.47e-05, | |
| "loss": 1.2321, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2706185567010309, | |
| "grad_norm": 0.9412080645561218, | |
| "learning_rate": 7.845e-05, | |
| "loss": 0.9694, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.28350515463917525, | |
| "grad_norm": 1.3718018531799316, | |
| "learning_rate": 8.22e-05, | |
| "loss": 0.9061, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2963917525773196, | |
| "grad_norm": 0.7417821884155273, | |
| "learning_rate": 8.594999999999999e-05, | |
| "loss": 0.8102, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.30927835051546393, | |
| "grad_norm": 0.9690466523170471, | |
| "learning_rate": 8.969999999999998e-05, | |
| "loss": 0.7434, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.32216494845360827, | |
| "grad_norm": 0.8293886780738831, | |
| "learning_rate": 9.345e-05, | |
| "loss": 0.7381, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.33505154639175255, | |
| "grad_norm": 0.8486959934234619, | |
| "learning_rate": 9.719999999999999e-05, | |
| "loss": 0.6224, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3479381443298969, | |
| "grad_norm": 0.7055938243865967, | |
| "learning_rate": 0.00010095, | |
| "loss": 0.6925, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.36082474226804123, | |
| "grad_norm": 0.7605512142181396, | |
| "learning_rate": 0.00010469999999999998, | |
| "loss": 0.6136, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.37371134020618557, | |
| "grad_norm": 0.6695568561553955, | |
| "learning_rate": 0.00010845, | |
| "loss": 0.6419, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.3865979381443299, | |
| "grad_norm": 0.8840702772140503, | |
| "learning_rate": 0.00011219999999999999, | |
| "loss": 0.6667, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.39948453608247425, | |
| "grad_norm": 0.7186158299446106, | |
| "learning_rate": 0.00011595, | |
| "loss": 0.6397, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.41237113402061853, | |
| "grad_norm": 0.6683000922203064, | |
| "learning_rate": 0.0001197, | |
| "loss": 0.5647, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4252577319587629, | |
| "grad_norm": 0.8537669777870178, | |
| "learning_rate": 0.00012345, | |
| "loss": 0.5624, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.4381443298969072, | |
| "grad_norm": 1.0267616510391235, | |
| "learning_rate": 0.00012719999999999997, | |
| "loss": 0.624, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.45103092783505155, | |
| "grad_norm": 0.6453070640563965, | |
| "learning_rate": 0.00013094999999999998, | |
| "loss": 0.5539, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.4639175257731959, | |
| "grad_norm": 1.0308513641357422, | |
| "learning_rate": 0.0001347, | |
| "loss": 0.5728, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.47680412371134023, | |
| "grad_norm": 0.560512125492096, | |
| "learning_rate": 0.00013845, | |
| "loss": 0.5305, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.4896907216494845, | |
| "grad_norm": 0.7196568250656128, | |
| "learning_rate": 0.0001422, | |
| "loss": 0.5669, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5025773195876289, | |
| "grad_norm": 0.5675578713417053, | |
| "learning_rate": 0.00014595, | |
| "loss": 0.5093, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.5154639175257731, | |
| "grad_norm": 0.8834163546562195, | |
| "learning_rate": 0.00014969999999999998, | |
| "loss": 0.5757, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5283505154639175, | |
| "grad_norm": 0.5083448886871338, | |
| "learning_rate": 0.00015344999999999996, | |
| "loss": 0.5412, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.5412371134020618, | |
| "grad_norm": 0.747595489025116, | |
| "learning_rate": 0.0001572, | |
| "loss": 0.5849, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.5541237113402062, | |
| "grad_norm": 0.5015640258789062, | |
| "learning_rate": 0.00016094999999999998, | |
| "loss": 0.5049, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.5670103092783505, | |
| "grad_norm": 0.7311388850212097, | |
| "learning_rate": 0.0001647, | |
| "loss": 0.4726, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5798969072164949, | |
| "grad_norm": 0.6942028999328613, | |
| "learning_rate": 0.00016844999999999997, | |
| "loss": 0.4934, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.5927835051546392, | |
| "grad_norm": 0.7268182635307312, | |
| "learning_rate": 0.00017219999999999998, | |
| "loss": 0.5074, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.6056701030927835, | |
| "grad_norm": 0.5172975063323975, | |
| "learning_rate": 0.00017595, | |
| "loss": 0.5293, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.6185567010309279, | |
| "grad_norm": 0.5973320603370667, | |
| "learning_rate": 0.00017969999999999998, | |
| "loss": 0.5018, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6314432989690721, | |
| "grad_norm": 0.6601810455322266, | |
| "learning_rate": 0.00018345, | |
| "loss": 0.4953, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.6443298969072165, | |
| "grad_norm": 0.7992896437644958, | |
| "learning_rate": 0.0001872, | |
| "loss": 0.4748, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.6572164948453608, | |
| "grad_norm": 0.5774939060211182, | |
| "learning_rate": 0.00019094999999999998, | |
| "loss": 0.4659, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.6701030927835051, | |
| "grad_norm": 0.7045702338218689, | |
| "learning_rate": 0.0001947, | |
| "loss": 0.4433, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6829896907216495, | |
| "grad_norm": 0.6067873239517212, | |
| "learning_rate": 0.00019844999999999997, | |
| "loss": 0.4872, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.6958762886597938, | |
| "grad_norm": 1.0101662874221802, | |
| "learning_rate": 0.0002022, | |
| "loss": 0.4933, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7087628865979382, | |
| "grad_norm": 0.4532999098300934, | |
| "learning_rate": 0.00020595, | |
| "loss": 0.474, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.7216494845360825, | |
| "grad_norm": 0.8769963383674622, | |
| "learning_rate": 0.00020969999999999997, | |
| "loss": 0.4923, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.7345360824742269, | |
| "grad_norm": 0.38735705614089966, | |
| "learning_rate": 0.00021344999999999998, | |
| "loss": 0.5358, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.7474226804123711, | |
| "grad_norm": 0.5298680663108826, | |
| "learning_rate": 0.00021719999999999997, | |
| "loss": 0.4961, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.7603092783505154, | |
| "grad_norm": 0.6393166780471802, | |
| "learning_rate": 0.00022095, | |
| "loss": 0.5057, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.7731958762886598, | |
| "grad_norm": 1.908589243888855, | |
| "learning_rate": 0.0002247, | |
| "loss": 0.5043, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7860824742268041, | |
| "grad_norm": 0.5039921402931213, | |
| "learning_rate": 0.00022844999999999997, | |
| "loss": 0.4767, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.7989690721649485, | |
| "grad_norm": 0.5750266909599304, | |
| "learning_rate": 0.00023219999999999998, | |
| "loss": 0.5002, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.8118556701030928, | |
| "grad_norm": 0.5906339883804321, | |
| "learning_rate": 0.00023594999999999996, | |
| "loss": 0.4776, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.8247422680412371, | |
| "grad_norm": 0.5718595385551453, | |
| "learning_rate": 0.0002397, | |
| "loss": 0.4934, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8376288659793815, | |
| "grad_norm": 0.5280390381813049, | |
| "learning_rate": 0.00024344999999999998, | |
| "loss": 0.4533, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.8505154639175257, | |
| "grad_norm": 1.016766905784607, | |
| "learning_rate": 0.0002472, | |
| "loss": 0.4795, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.8634020618556701, | |
| "grad_norm": 0.4700835943222046, | |
| "learning_rate": 0.00025095, | |
| "loss": 0.4399, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.8762886597938144, | |
| "grad_norm": 0.7027618885040283, | |
| "learning_rate": 0.00025469999999999996, | |
| "loss": 0.4801, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8891752577319587, | |
| "grad_norm": 0.41857969760894775, | |
| "learning_rate": 0.00025844999999999997, | |
| "loss": 0.4449, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.9020618556701031, | |
| "grad_norm": 0.7206704020500183, | |
| "learning_rate": 0.0002622, | |
| "loss": 0.4648, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.9149484536082474, | |
| "grad_norm": 0.485895037651062, | |
| "learning_rate": 0.00026595, | |
| "loss": 0.4598, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.9278350515463918, | |
| "grad_norm": 0.7267158031463623, | |
| "learning_rate": 0.0002697, | |
| "loss": 0.462, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9407216494845361, | |
| "grad_norm": 0.4129009246826172, | |
| "learning_rate": 0.00027344999999999995, | |
| "loss": 0.4585, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.9536082474226805, | |
| "grad_norm": 0.6467506885528564, | |
| "learning_rate": 0.0002772, | |
| "loss": 0.4539, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.9664948453608248, | |
| "grad_norm": 0.6241980791091919, | |
| "learning_rate": 0.00028094999999999997, | |
| "loss": 0.4917, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.979381443298969, | |
| "grad_norm": 0.817642867565155, | |
| "learning_rate": 0.0002847, | |
| "loss": 0.4337, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9922680412371134, | |
| "grad_norm": 1.18275785446167, | |
| "learning_rate": 0.00028845, | |
| "loss": 0.4591, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.0051546391752577, | |
| "grad_norm": 0.5338313579559326, | |
| "learning_rate": 0.00029219999999999995, | |
| "loss": 0.5187, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.018041237113402, | |
| "grad_norm": 0.40176838636398315, | |
| "learning_rate": 0.00029595, | |
| "loss": 0.4012, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.0309278350515463, | |
| "grad_norm": 0.6469115018844604, | |
| "learning_rate": 0.00029969999999999997, | |
| "loss": 0.4752, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0309278350515463, | |
| "eval_loss": 0.59544438123703, | |
| "eval_runtime": 196.5387, | |
| "eval_samples_per_second": 6.233, | |
| "eval_steps_per_second": 0.784, | |
| "eval_wer": 1.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0438144329896908, | |
| "grad_norm": 0.6261550188064575, | |
| "learning_rate": 0.00029987722419928825, | |
| "loss": 0.4102, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.056701030927835, | |
| "grad_norm": 0.8379774689674377, | |
| "learning_rate": 0.00029974377224199284, | |
| "loss": 0.4658, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.0695876288659794, | |
| "grad_norm": 0.503239631652832, | |
| "learning_rate": 0.0002996103202846975, | |
| "loss": 0.4167, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.0824742268041236, | |
| "grad_norm": 0.3943336606025696, | |
| "learning_rate": 0.00029947686832740213, | |
| "loss": 0.4686, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.0953608247422681, | |
| "grad_norm": 0.4782889783382416, | |
| "learning_rate": 0.0002993434163701067, | |
| "loss": 0.4003, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.1082474226804124, | |
| "grad_norm": 0.37270641326904297, | |
| "learning_rate": 0.00029920996441281137, | |
| "loss": 0.4578, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.1211340206185567, | |
| "grad_norm": 0.5629658699035645, | |
| "learning_rate": 0.000299076512455516, | |
| "loss": 0.3687, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.134020618556701, | |
| "grad_norm": 0.3787396252155304, | |
| "learning_rate": 0.0002989430604982206, | |
| "loss": 0.4322, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.1469072164948453, | |
| "grad_norm": 0.6377401947975159, | |
| "learning_rate": 0.00029880960854092526, | |
| "loss": 0.4403, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.1597938144329896, | |
| "grad_norm": 0.37100082635879517, | |
| "learning_rate": 0.00029867615658362985, | |
| "loss": 0.453, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.172680412371134, | |
| "grad_norm": 0.7276923060417175, | |
| "learning_rate": 0.0002985427046263345, | |
| "loss": 0.4145, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.1855670103092784, | |
| "grad_norm": 0.8605408668518066, | |
| "learning_rate": 0.00029840925266903914, | |
| "loss": 0.4539, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.1984536082474226, | |
| "grad_norm": 0.45449143648147583, | |
| "learning_rate": 0.00029827580071174373, | |
| "loss": 0.4031, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.211340206185567, | |
| "grad_norm": 0.36433449387550354, | |
| "learning_rate": 0.0002981423487544484, | |
| "loss": 0.4548, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.2242268041237114, | |
| "grad_norm": 0.48323678970336914, | |
| "learning_rate": 0.000298008896797153, | |
| "loss": 0.4236, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.2371134020618557, | |
| "grad_norm": 0.6236255764961243, | |
| "learning_rate": 0.0002978754448398576, | |
| "loss": 0.438, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.721978485584259, | |
| "learning_rate": 0.00029774199288256227, | |
| "loss": 0.3803, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.2628865979381443, | |
| "grad_norm": 0.33121246099472046, | |
| "learning_rate": 0.00029760854092526686, | |
| "loss": 0.4759, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.2757731958762886, | |
| "grad_norm": 0.6318331360816956, | |
| "learning_rate": 0.0002974750889679715, | |
| "loss": 0.3616, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.2886597938144329, | |
| "grad_norm": 0.30613037943840027, | |
| "learning_rate": 0.0002973416370106761, | |
| "loss": 0.4355, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.3015463917525774, | |
| "grad_norm": 0.6804624199867249, | |
| "learning_rate": 0.00029720818505338075, | |
| "loss": 0.4296, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.3144329896907216, | |
| "grad_norm": 0.5847098231315613, | |
| "learning_rate": 0.0002970747330960854, | |
| "loss": 0.4774, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.327319587628866, | |
| "grad_norm": 0.43900150060653687, | |
| "learning_rate": 0.00029694128113879, | |
| "loss": 0.4249, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.3402061855670104, | |
| "grad_norm": 0.26748043298721313, | |
| "learning_rate": 0.00029680782918149463, | |
| "loss": 0.4713, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.3530927835051547, | |
| "grad_norm": 0.5893319845199585, | |
| "learning_rate": 0.0002966743772241993, | |
| "loss": 0.3566, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.365979381443299, | |
| "grad_norm": 0.2939490079879761, | |
| "learning_rate": 0.00029654092526690387, | |
| "loss": 0.427, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.3788659793814433, | |
| "grad_norm": 0.6302582025527954, | |
| "learning_rate": 0.0002964074733096085, | |
| "loss": 0.413, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.3917525773195876, | |
| "grad_norm": 0.5481074452400208, | |
| "learning_rate": 0.0002962740213523131, | |
| "loss": 0.5027, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.4046391752577319, | |
| "grad_norm": 0.41078057885169983, | |
| "learning_rate": 0.00029614056939501776, | |
| "loss": 0.3935, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.4175257731958764, | |
| "grad_norm": 0.5955342054367065, | |
| "learning_rate": 0.0002960071174377224, | |
| "loss": 0.4454, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.4304123711340206, | |
| "grad_norm": 0.47366246581077576, | |
| "learning_rate": 0.000295873665480427, | |
| "loss": 0.3861, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.443298969072165, | |
| "grad_norm": 0.5429800748825073, | |
| "learning_rate": 0.00029574021352313164, | |
| "loss": 0.4525, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.4561855670103092, | |
| "grad_norm": 0.5089389681816101, | |
| "learning_rate": 0.00029560676156583623, | |
| "loss": 0.3727, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.4690721649484537, | |
| "grad_norm": 0.3353477716445923, | |
| "learning_rate": 0.0002954733096085409, | |
| "loss": 0.4855, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.481958762886598, | |
| "grad_norm": 0.6160407662391663, | |
| "learning_rate": 0.0002953398576512455, | |
| "loss": 0.3635, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.4948453608247423, | |
| "grad_norm": 0.3237319886684418, | |
| "learning_rate": 0.0002952064056939501, | |
| "loss": 0.4285, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.5077319587628866, | |
| "grad_norm": 0.450309693813324, | |
| "learning_rate": 0.00029507295373665477, | |
| "loss": 0.3759, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.5206185567010309, | |
| "grad_norm": 0.3550674319267273, | |
| "learning_rate": 0.0002949395017793594, | |
| "loss": 0.4442, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.5335051546391751, | |
| "grad_norm": 0.3989160656929016, | |
| "learning_rate": 0.000294806049822064, | |
| "loss": 0.4255, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.5463917525773194, | |
| "grad_norm": 0.5774266719818115, | |
| "learning_rate": 0.00029467259786476865, | |
| "loss": 0.4526, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.559278350515464, | |
| "grad_norm": 0.6264383792877197, | |
| "learning_rate": 0.0002945391459074733, | |
| "loss": 0.4097, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.5721649484536082, | |
| "grad_norm": 0.33969295024871826, | |
| "learning_rate": 0.0002944056939501779, | |
| "loss": 0.4241, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.5850515463917527, | |
| "grad_norm": 0.39299994707107544, | |
| "learning_rate": 0.00029427224199288254, | |
| "loss": 0.3778, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.597938144329897, | |
| "grad_norm": 0.42388004064559937, | |
| "learning_rate": 0.0002941387900355872, | |
| "loss": 0.4004, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.6108247422680413, | |
| "grad_norm": 0.9131516814231873, | |
| "learning_rate": 0.0002940053380782918, | |
| "loss": 0.3531, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.6237113402061856, | |
| "grad_norm": 0.296908974647522, | |
| "learning_rate": 0.0002938718861209964, | |
| "loss": 0.4613, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.6365979381443299, | |
| "grad_norm": 0.4583122134208679, | |
| "learning_rate": 0.000293738434163701, | |
| "loss": 0.3833, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.6494845360824741, | |
| "grad_norm": 0.35052600502967834, | |
| "learning_rate": 0.00029360498220640566, | |
| "loss": 0.4432, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.6623711340206184, | |
| "grad_norm": 0.532720685005188, | |
| "learning_rate": 0.0002934715302491103, | |
| "loss": 0.3635, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.675257731958763, | |
| "grad_norm": 0.3807854652404785, | |
| "learning_rate": 0.0002933380782918149, | |
| "loss": 0.43, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.6881443298969072, | |
| "grad_norm": 0.5288501381874084, | |
| "learning_rate": 0.00029320462633451955, | |
| "loss": 0.3449, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.7010309278350515, | |
| "grad_norm": 0.3881712853908539, | |
| "learning_rate": 0.0002930711743772242, | |
| "loss": 0.4406, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.713917525773196, | |
| "grad_norm": 0.42132484912872314, | |
| "learning_rate": 0.0002929377224199288, | |
| "loss": 0.3888, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.7268041237113403, | |
| "grad_norm": 0.3974430561065674, | |
| "learning_rate": 0.00029280427046263343, | |
| "loss": 0.4265, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.7396907216494846, | |
| "grad_norm": 0.4618494212627411, | |
| "learning_rate": 0.0002926708185053381, | |
| "loss": 0.3537, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.7525773195876289, | |
| "grad_norm": 0.347777783870697, | |
| "learning_rate": 0.00029253736654804267, | |
| "loss": 0.4282, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.7654639175257731, | |
| "grad_norm": 0.5885565280914307, | |
| "learning_rate": 0.0002924039145907473, | |
| "loss": 0.3694, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.7783505154639174, | |
| "grad_norm": 0.5113171935081482, | |
| "learning_rate": 0.00029227046263345197, | |
| "loss": 0.5256, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.7912371134020617, | |
| "grad_norm": 0.7724625468254089, | |
| "learning_rate": 0.00029213701067615656, | |
| "loss": 0.3649, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.8041237113402062, | |
| "grad_norm": 0.44778281450271606, | |
| "learning_rate": 0.0002920035587188612, | |
| "loss": 0.4313, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.8170103092783505, | |
| "grad_norm": 0.8251272439956665, | |
| "learning_rate": 0.00029187010676156585, | |
| "loss": 0.3839, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.829896907216495, | |
| "grad_norm": 0.4858299493789673, | |
| "learning_rate": 0.00029173665480427044, | |
| "loss": 0.4489, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.8427835051546393, | |
| "grad_norm": 0.5172144770622253, | |
| "learning_rate": 0.0002916032028469751, | |
| "loss": 0.409, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.8556701030927836, | |
| "grad_norm": 0.34371522068977356, | |
| "learning_rate": 0.0002914697508896797, | |
| "loss": 0.4365, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.8685567010309279, | |
| "grad_norm": 0.5957440137863159, | |
| "learning_rate": 0.00029133629893238433, | |
| "loss": 0.3611, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.8814432989690721, | |
| "grad_norm": 0.3856901526451111, | |
| "learning_rate": 0.000291202846975089, | |
| "loss": 0.451, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.8943298969072164, | |
| "grad_norm": 0.5961311459541321, | |
| "learning_rate": 0.00029106939501779357, | |
| "loss": 0.3934, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.9072164948453607, | |
| "grad_norm": 0.4491939842700958, | |
| "learning_rate": 0.0002909359430604982, | |
| "loss": 0.4466, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.920103092783505, | |
| "grad_norm": 0.41777607798576355, | |
| "learning_rate": 0.0002908024911032028, | |
| "loss": 0.3754, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.9329896907216495, | |
| "grad_norm": 0.42550450563430786, | |
| "learning_rate": 0.00029066903914590745, | |
| "loss": 0.4027, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.9458762886597938, | |
| "grad_norm": 0.427433043718338, | |
| "learning_rate": 0.0002905355871886121, | |
| "loss": 0.3603, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.9587628865979383, | |
| "grad_norm": 0.8279537558555603, | |
| "learning_rate": 0.0002904021352313167, | |
| "loss": 0.4098, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.9716494845360826, | |
| "grad_norm": 0.6759387850761414, | |
| "learning_rate": 0.00029026868327402134, | |
| "loss": 0.3756, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.9845360824742269, | |
| "grad_norm": 0.29660704731941223, | |
| "learning_rate": 0.000290135231316726, | |
| "loss": 0.4591, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.9974226804123711, | |
| "grad_norm": 0.4846726655960083, | |
| "learning_rate": 0.0002900017793594306, | |
| "loss": 0.3944, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.0103092783505154, | |
| "grad_norm": 0.4863591492176056, | |
| "learning_rate": 0.0002898683274021352, | |
| "loss": 0.3813, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.0231958762886597, | |
| "grad_norm": 0.37275585532188416, | |
| "learning_rate": 0.0002897348754448398, | |
| "loss": 0.3753, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.036082474226804, | |
| "grad_norm": 0.36628881096839905, | |
| "learning_rate": 0.00028960142348754446, | |
| "loss": 0.403, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.0489690721649483, | |
| "grad_norm": 0.4523802697658539, | |
| "learning_rate": 0.0002894679715302491, | |
| "loss": 0.3601, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.0618556701030926, | |
| "grad_norm": 0.4722582697868347, | |
| "learning_rate": 0.0002893345195729537, | |
| "loss": 0.3858, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0618556701030926, | |
| "eval_loss": 0.46308374404907227, | |
| "eval_runtime": 201.4429, | |
| "eval_samples_per_second": 6.081, | |
| "eval_steps_per_second": 0.764, | |
| "eval_wer": 1.0057142857142858, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0747422680412373, | |
| "grad_norm": 0.5041220784187317, | |
| "learning_rate": 0.00028920106761565835, | |
| "loss": 0.3518, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.0876288659793816, | |
| "grad_norm": 0.48610439896583557, | |
| "learning_rate": 0.00028906761565836294, | |
| "loss": 0.3818, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.100515463917526, | |
| "grad_norm": 0.6834833025932312, | |
| "learning_rate": 0.0002889341637010676, | |
| "loss": 0.3778, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.11340206185567, | |
| "grad_norm": 0.687245786190033, | |
| "learning_rate": 0.00028880071174377224, | |
| "loss": 0.3771, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.1262886597938144, | |
| "grad_norm": 0.706832230091095, | |
| "learning_rate": 0.00028866725978647683, | |
| "loss": 0.3868, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.1391752577319587, | |
| "grad_norm": 0.7049499154090881, | |
| "learning_rate": 0.0002885338078291815, | |
| "loss": 0.3646, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.152061855670103, | |
| "grad_norm": 0.40853607654571533, | |
| "learning_rate": 0.00028840035587188607, | |
| "loss": 0.3235, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.1649484536082473, | |
| "grad_norm": 0.3675331771373749, | |
| "learning_rate": 0.0002882669039145907, | |
| "loss": 0.4079, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.1778350515463916, | |
| "grad_norm": 1.3320142030715942, | |
| "learning_rate": 0.00028813345195729536, | |
| "loss": 0.3031, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.1907216494845363, | |
| "grad_norm": 0.6416336297988892, | |
| "learning_rate": 0.00028799999999999995, | |
| "loss": 0.4087, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.2036082474226806, | |
| "grad_norm": 0.43675994873046875, | |
| "learning_rate": 0.0002878665480427046, | |
| "loss": 0.3577, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.216494845360825, | |
| "grad_norm": 0.3541963994503021, | |
| "learning_rate": 0.00028773309608540925, | |
| "loss": 0.3706, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.229381443298969, | |
| "grad_norm": 0.4469320476055145, | |
| "learning_rate": 0.00028759964412811384, | |
| "loss": 0.3692, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.2422680412371134, | |
| "grad_norm": 0.4056352376937866, | |
| "learning_rate": 0.0002874661921708185, | |
| "loss": 0.3659, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.2551546391752577, | |
| "grad_norm": 0.3547820746898651, | |
| "learning_rate": 0.0002873327402135231, | |
| "loss": 0.3564, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.268041237113402, | |
| "grad_norm": 0.32645678520202637, | |
| "learning_rate": 0.0002871992882562277, | |
| "loss": 0.3346, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.2809278350515463, | |
| "grad_norm": 0.7449667453765869, | |
| "learning_rate": 0.00028706583629893237, | |
| "loss": 0.3696, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.2938144329896906, | |
| "grad_norm": 0.4612623155117035, | |
| "learning_rate": 0.00028693238434163696, | |
| "loss": 0.3925, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.306701030927835, | |
| "grad_norm": 0.6325463056564331, | |
| "learning_rate": 0.0002867989323843416, | |
| "loss": 0.3512, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.319587628865979, | |
| "grad_norm": 0.29471156001091003, | |
| "learning_rate": 0.0002866654804270462, | |
| "loss": 0.3569, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.332474226804124, | |
| "grad_norm": 1.076217770576477, | |
| "learning_rate": 0.00028653202846975085, | |
| "loss": 0.3375, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.345360824742268, | |
| "grad_norm": 0.614162027835846, | |
| "learning_rate": 0.0002863985765124555, | |
| "loss": 0.4078, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.3582474226804124, | |
| "grad_norm": 0.4514384865760803, | |
| "learning_rate": 0.0002862651245551601, | |
| "loss": 0.3758, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.3711340206185567, | |
| "grad_norm": 0.314336359500885, | |
| "learning_rate": 0.00028613167259786473, | |
| "loss": 0.3765, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.384020618556701, | |
| "grad_norm": 0.5032577514648438, | |
| "learning_rate": 0.0002859982206405694, | |
| "loss": 0.3226, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.3969072164948453, | |
| "grad_norm": 0.6402880549430847, | |
| "learning_rate": 0.000285864768683274, | |
| "loss": 0.3986, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.4097938144329896, | |
| "grad_norm": 0.5510269403457642, | |
| "learning_rate": 0.0002857313167259786, | |
| "loss": 0.3757, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.422680412371134, | |
| "grad_norm": 0.5594025254249573, | |
| "learning_rate": 0.0002855978647686832, | |
| "loss": 0.4297, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.4355670103092786, | |
| "grad_norm": 0.3621445894241333, | |
| "learning_rate": 0.00028546441281138786, | |
| "loss": 0.3336, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.448453608247423, | |
| "grad_norm": 0.5683943033218384, | |
| "learning_rate": 0.0002853309608540925, | |
| "loss": 0.3395, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.461340206185567, | |
| "grad_norm": 0.7661644816398621, | |
| "learning_rate": 0.0002851975088967971, | |
| "loss": 0.377, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.4742268041237114, | |
| "grad_norm": 0.456636905670166, | |
| "learning_rate": 0.00028506405693950175, | |
| "loss": 0.3645, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.4871134020618557, | |
| "grad_norm": 0.49251827597618103, | |
| "learning_rate": 0.0002849306049822064, | |
| "loss": 0.34, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.32308971881866455, | |
| "learning_rate": 0.000284797153024911, | |
| "loss": 0.4092, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.5128865979381443, | |
| "grad_norm": 0.9722476005554199, | |
| "learning_rate": 0.00028466370106761563, | |
| "loss": 0.3651, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.5257731958762886, | |
| "grad_norm": 0.4552549123764038, | |
| "learning_rate": 0.0002845302491103203, | |
| "loss": 0.3837, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.538659793814433, | |
| "grad_norm": 0.4148350656032562, | |
| "learning_rate": 0.00028439679715302487, | |
| "loss": 0.3931, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.551546391752577, | |
| "grad_norm": 0.41644537448883057, | |
| "learning_rate": 0.0002842633451957295, | |
| "loss": 0.3677, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.5644329896907214, | |
| "grad_norm": 0.3989255428314209, | |
| "learning_rate": 0.00028412989323843416, | |
| "loss": 0.3759, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.5773195876288657, | |
| "grad_norm": 0.472091943025589, | |
| "learning_rate": 0.00028399644128113876, | |
| "loss": 0.4008, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.5902061855670104, | |
| "grad_norm": 0.9913691282272339, | |
| "learning_rate": 0.0002838629893238434, | |
| "loss": 0.3555, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 2.6030927835051547, | |
| "grad_norm": 0.425589382648468, | |
| "learning_rate": 0.000283729537366548, | |
| "loss": 0.3774, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.615979381443299, | |
| "grad_norm": 0.7836791276931763, | |
| "learning_rate": 0.00028359608540925264, | |
| "loss": 0.317, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 2.6288659793814433, | |
| "grad_norm": 0.7071629166603088, | |
| "learning_rate": 0.0002834626334519573, | |
| "loss": 0.4039, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.6417525773195876, | |
| "grad_norm": 0.5756880640983582, | |
| "learning_rate": 0.0002833291814946619, | |
| "loss": 0.342, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.654639175257732, | |
| "grad_norm": 0.425029993057251, | |
| "learning_rate": 0.0002831957295373665, | |
| "loss": 0.3765, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.667525773195876, | |
| "grad_norm": 0.5212023854255676, | |
| "learning_rate": 0.0002830622775800712, | |
| "loss": 0.3681, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 2.680412371134021, | |
| "grad_norm": 0.39988288283348083, | |
| "learning_rate": 0.00028292882562277577, | |
| "loss": 0.3525, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.693298969072165, | |
| "grad_norm": 0.31254854798316956, | |
| "learning_rate": 0.0002827953736654804, | |
| "loss": 0.3521, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 2.7061855670103094, | |
| "grad_norm": 0.5564956665039062, | |
| "learning_rate": 0.00028266192170818506, | |
| "loss": 0.3602, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.7190721649484537, | |
| "grad_norm": 0.45266616344451904, | |
| "learning_rate": 0.00028252846975088965, | |
| "loss": 0.3462, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 2.731958762886598, | |
| "grad_norm": 0.26552554965019226, | |
| "learning_rate": 0.0002823950177935943, | |
| "loss": 0.3787, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.7448453608247423, | |
| "grad_norm": 0.5850217938423157, | |
| "learning_rate": 0.00028226156583629894, | |
| "loss": 0.3289, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 2.7577319587628866, | |
| "grad_norm": 0.47871604561805725, | |
| "learning_rate": 0.00028212811387900354, | |
| "loss": 0.4246, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.770618556701031, | |
| "grad_norm": 0.5951977372169495, | |
| "learning_rate": 0.0002819946619217082, | |
| "loss": 0.3326, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.783505154639175, | |
| "grad_norm": 0.5806294083595276, | |
| "learning_rate": 0.0002818612099644128, | |
| "loss": 0.3952, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.7963917525773194, | |
| "grad_norm": 0.6218500733375549, | |
| "learning_rate": 0.0002817277580071174, | |
| "loss": 0.3222, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 2.8092783505154637, | |
| "grad_norm": 0.38604310154914856, | |
| "learning_rate": 0.00028159430604982207, | |
| "loss": 0.4408, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.822164948453608, | |
| "grad_norm": 2.576910972595215, | |
| "learning_rate": 0.00028146085409252666, | |
| "loss": 0.355, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 2.8350515463917527, | |
| "grad_norm": 0.28045961260795593, | |
| "learning_rate": 0.0002813274021352313, | |
| "loss": 0.3479, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.847938144329897, | |
| "grad_norm": 0.6178203225135803, | |
| "learning_rate": 0.00028119395017793596, | |
| "loss": 0.3097, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 2.8608247422680413, | |
| "grad_norm": 0.36241769790649414, | |
| "learning_rate": 0.00028106049822064055, | |
| "loss": 0.3762, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.8737113402061856, | |
| "grad_norm": 0.6459288597106934, | |
| "learning_rate": 0.0002809270462633452, | |
| "loss": 0.3261, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 2.88659793814433, | |
| "grad_norm": 0.458281546831131, | |
| "learning_rate": 0.0002807935943060498, | |
| "loss": 0.378, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.899484536082474, | |
| "grad_norm": 0.4269348382949829, | |
| "learning_rate": 0.00028066014234875443, | |
| "loss": 0.3483, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.9123711340206184, | |
| "grad_norm": 0.36018800735473633, | |
| "learning_rate": 0.0002805266903914591, | |
| "loss": 0.3592, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.925257731958763, | |
| "grad_norm": 0.6075245141983032, | |
| "learning_rate": 0.00028039323843416367, | |
| "loss": 0.3, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 2.9381443298969074, | |
| "grad_norm": 0.3082279860973358, | |
| "learning_rate": 0.0002802597864768683, | |
| "loss": 0.4306, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.9510309278350517, | |
| "grad_norm": 0.6344396471977234, | |
| "learning_rate": 0.0002801263345195729, | |
| "loss": 0.3795, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 2.963917525773196, | |
| "grad_norm": 0.3072182536125183, | |
| "learning_rate": 0.00027999288256227756, | |
| "loss": 0.3425, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.9768041237113403, | |
| "grad_norm": 0.5076513886451721, | |
| "learning_rate": 0.0002798594306049822, | |
| "loss": 0.3214, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.9896907216494846, | |
| "grad_norm": 0.34852054715156555, | |
| "learning_rate": 0.0002797259786476868, | |
| "loss": 0.3678, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.002577319587629, | |
| "grad_norm": 0.36334386467933655, | |
| "learning_rate": 0.00027959252669039144, | |
| "loss": 0.3984, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 3.015463917525773, | |
| "grad_norm": 0.46211138367652893, | |
| "learning_rate": 0.00027945907473309604, | |
| "loss": 0.2912, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.0283505154639174, | |
| "grad_norm": 0.3730103373527527, | |
| "learning_rate": 0.0002793256227758007, | |
| "loss": 0.337, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 3.0412371134020617, | |
| "grad_norm": 0.37066757678985596, | |
| "learning_rate": 0.00027919217081850533, | |
| "loss": 0.2959, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.054123711340206, | |
| "grad_norm": 0.5712897181510925, | |
| "learning_rate": 0.0002790587188612099, | |
| "loss": 0.3279, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 3.0670103092783507, | |
| "grad_norm": 0.39846915006637573, | |
| "learning_rate": 0.00027892526690391457, | |
| "loss": 0.3532, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 3.079896907216495, | |
| "grad_norm": 0.3366047143936157, | |
| "learning_rate": 0.0002787918149466192, | |
| "loss": 0.3486, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 3.0927835051546393, | |
| "grad_norm": 0.5053852796554565, | |
| "learning_rate": 0.0002786583629893238, | |
| "loss": 0.3571, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.0927835051546393, | |
| "eval_loss": 0.48639050126075745, | |
| "eval_runtime": 195.6249, | |
| "eval_samples_per_second": 6.262, | |
| "eval_steps_per_second": 0.787, | |
| "eval_wer": 1.0073469387755103, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.1056701030927836, | |
| "grad_norm": 0.35252827405929565, | |
| "learning_rate": 0.00027852491103202845, | |
| "loss": 0.3401, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 3.118556701030928, | |
| "grad_norm": 0.3530094027519226, | |
| "learning_rate": 0.00027839145907473305, | |
| "loss": 0.3094, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 3.131443298969072, | |
| "grad_norm": 0.9595320224761963, | |
| "learning_rate": 0.0002782580071174377, | |
| "loss": 0.3692, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 3.1443298969072164, | |
| "grad_norm": 0.3585176467895508, | |
| "learning_rate": 0.00027812455516014234, | |
| "loss": 0.3351, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.1572164948453607, | |
| "grad_norm": 0.4432651996612549, | |
| "learning_rate": 0.00027799110320284693, | |
| "loss": 0.3494, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 3.170103092783505, | |
| "grad_norm": 0.5367820858955383, | |
| "learning_rate": 0.0002778576512455516, | |
| "loss": 0.3311, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.1829896907216493, | |
| "grad_norm": 0.3430980443954468, | |
| "learning_rate": 0.00027772419928825617, | |
| "loss": 0.3357, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 3.195876288659794, | |
| "grad_norm": 1.0889408588409424, | |
| "learning_rate": 0.0002775907473309608, | |
| "loss": 0.3056, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.2087628865979383, | |
| "grad_norm": 0.3883308172225952, | |
| "learning_rate": 0.00027745729537366546, | |
| "loss": 0.3251, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 3.2216494845360826, | |
| "grad_norm": 0.41912856698036194, | |
| "learning_rate": 0.00027732384341637006, | |
| "loss": 0.3292, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.234536082474227, | |
| "grad_norm": 0.45084699988365173, | |
| "learning_rate": 0.0002771903914590747, | |
| "loss": 0.3516, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 3.247422680412371, | |
| "grad_norm": 0.4294661581516266, | |
| "learning_rate": 0.00027705693950177935, | |
| "loss": 0.3297, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.2603092783505154, | |
| "grad_norm": 0.25462472438812256, | |
| "learning_rate": 0.00027692348754448394, | |
| "loss": 0.3513, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 3.2731958762886597, | |
| "grad_norm": 0.2990482747554779, | |
| "learning_rate": 0.0002767900355871886, | |
| "loss": 0.3472, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.286082474226804, | |
| "grad_norm": 0.5474823713302612, | |
| "learning_rate": 0.0002766565836298932, | |
| "loss": 0.3404, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 3.2989690721649483, | |
| "grad_norm": 0.31416311860084534, | |
| "learning_rate": 0.00027652313167259783, | |
| "loss": 0.3052, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.3118556701030926, | |
| "grad_norm": 0.6990143656730652, | |
| "learning_rate": 0.0002763896797153025, | |
| "loss": 0.3661, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 3.3247422680412373, | |
| "grad_norm": 0.4336131811141968, | |
| "learning_rate": 0.00027625622775800707, | |
| "loss": 0.2757, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.3376288659793816, | |
| "grad_norm": 0.35333672165870667, | |
| "learning_rate": 0.0002761227758007117, | |
| "loss": 0.3275, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 3.350515463917526, | |
| "grad_norm": 0.7743633985519409, | |
| "learning_rate": 0.0002759893238434163, | |
| "loss": 0.2855, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.36340206185567, | |
| "grad_norm": 0.5316669344902039, | |
| "learning_rate": 0.00027585587188612095, | |
| "loss": 0.3742, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 3.3762886597938144, | |
| "grad_norm": 0.3642534017562866, | |
| "learning_rate": 0.0002757277580071174, | |
| "loss": 0.3676, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.3891752577319587, | |
| "grad_norm": 0.6164928078651428, | |
| "learning_rate": 0.00027559430604982203, | |
| "loss": 0.3658, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 3.402061855670103, | |
| "grad_norm": 0.3114078938961029, | |
| "learning_rate": 0.0002754608540925266, | |
| "loss": 0.2681, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.4149484536082473, | |
| "grad_norm": 0.8610777258872986, | |
| "learning_rate": 0.00027532740213523127, | |
| "loss": 0.3607, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 3.4278350515463916, | |
| "grad_norm": 0.49727797508239746, | |
| "learning_rate": 0.0002751939501779359, | |
| "loss": 0.3306, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.4407216494845363, | |
| "grad_norm": 0.3058234751224518, | |
| "learning_rate": 0.0002750604982206405, | |
| "loss": 0.3313, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 3.4536082474226806, | |
| "grad_norm": 0.5317339301109314, | |
| "learning_rate": 0.00027492704626334516, | |
| "loss": 0.3316, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.466494845360825, | |
| "grad_norm": 0.4639209806919098, | |
| "learning_rate": 0.0002747935943060498, | |
| "loss": 0.3773, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 3.479381443298969, | |
| "grad_norm": 0.28237494826316833, | |
| "learning_rate": 0.0002746601423487544, | |
| "loss": 0.3275, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.4922680412371134, | |
| "grad_norm": 0.44046372175216675, | |
| "learning_rate": 0.00027452669039145904, | |
| "loss": 0.3777, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 3.5051546391752577, | |
| "grad_norm": 0.40542760491371155, | |
| "learning_rate": 0.0002743932384341637, | |
| "loss": 0.3197, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.518041237113402, | |
| "grad_norm": 0.587062418460846, | |
| "learning_rate": 0.0002742597864768683, | |
| "loss": 0.3441, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 3.5309278350515463, | |
| "grad_norm": 0.34003278613090515, | |
| "learning_rate": 0.00027412633451957293, | |
| "loss": 0.3189, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.5438144329896906, | |
| "grad_norm": 0.40320533514022827, | |
| "learning_rate": 0.0002739928825622776, | |
| "loss": 0.3115, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 3.556701030927835, | |
| "grad_norm": 0.34437236189842224, | |
| "learning_rate": 0.00027385943060498217, | |
| "loss": 0.3344, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.569587628865979, | |
| "grad_norm": 0.42826181650161743, | |
| "learning_rate": 0.0002737259786476868, | |
| "loss": 0.3716, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 3.582474226804124, | |
| "grad_norm": 0.4200308620929718, | |
| "learning_rate": 0.00027359252669039146, | |
| "loss": 0.3511, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 3.595360824742268, | |
| "grad_norm": 0.3897481858730316, | |
| "learning_rate": 0.00027345907473309605, | |
| "loss": 0.3651, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 3.6082474226804124, | |
| "grad_norm": 0.3066980838775635, | |
| "learning_rate": 0.0002733256227758007, | |
| "loss": 0.314, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.6211340206185567, | |
| "grad_norm": 0.39002224802970886, | |
| "learning_rate": 0.0002731921708185053, | |
| "loss": 0.411, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 3.634020618556701, | |
| "grad_norm": 0.8328560590744019, | |
| "learning_rate": 0.00027305871886120994, | |
| "loss": 0.3154, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 3.6469072164948453, | |
| "grad_norm": 0.43831050395965576, | |
| "learning_rate": 0.0002729252669039146, | |
| "loss": 0.396, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 3.6597938144329896, | |
| "grad_norm": 0.3992260694503784, | |
| "learning_rate": 0.0002727918149466192, | |
| "loss": 0.3142, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.6726804123711343, | |
| "grad_norm": 0.2941119074821472, | |
| "learning_rate": 0.0002726583629893238, | |
| "loss": 0.3527, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 3.6855670103092786, | |
| "grad_norm": 0.4881301522254944, | |
| "learning_rate": 0.00027252491103202847, | |
| "loss": 0.3171, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 3.698453608247423, | |
| "grad_norm": 0.3898142874240875, | |
| "learning_rate": 0.00027239145907473306, | |
| "loss": 0.3817, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 3.711340206185567, | |
| "grad_norm": 0.31580379605293274, | |
| "learning_rate": 0.0002722580071174377, | |
| "loss": 0.2595, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.7242268041237114, | |
| "grad_norm": 0.3299170136451721, | |
| "learning_rate": 0.00027212455516014236, | |
| "loss": 0.3416, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 3.7371134020618557, | |
| "grad_norm": 0.6964473724365234, | |
| "learning_rate": 0.00027199110320284695, | |
| "loss": 0.2942, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 0.3565264046192169, | |
| "learning_rate": 0.0002718576512455516, | |
| "loss": 0.3283, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 3.7628865979381443, | |
| "grad_norm": 0.28134214878082275, | |
| "learning_rate": 0.00027172419928825624, | |
| "loss": 0.3203, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.7757731958762886, | |
| "grad_norm": 0.4196174442768097, | |
| "learning_rate": 0.00027159074733096083, | |
| "loss": 0.3396, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 3.788659793814433, | |
| "grad_norm": 0.4770311415195465, | |
| "learning_rate": 0.0002714572953736655, | |
| "loss": 0.3144, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.801546391752577, | |
| "grad_norm": 0.29138994216918945, | |
| "learning_rate": 0.0002713238434163701, | |
| "loss": 0.3403, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 3.8144329896907214, | |
| "grad_norm": 0.47035351395606995, | |
| "learning_rate": 0.0002711903914590747, | |
| "loss": 0.2623, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.8273195876288657, | |
| "grad_norm": 0.32780078053474426, | |
| "learning_rate": 0.00027105693950177937, | |
| "loss": 0.3863, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 3.8402061855670104, | |
| "grad_norm": 0.4419979751110077, | |
| "learning_rate": 0.00027092348754448396, | |
| "loss": 0.3207, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.8530927835051547, | |
| "grad_norm": 0.28167861700057983, | |
| "learning_rate": 0.0002707900355871886, | |
| "loss": 0.3308, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 3.865979381443299, | |
| "grad_norm": 0.3411625921726227, | |
| "learning_rate": 0.00027065658362989325, | |
| "loss": 0.3387, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.8788659793814433, | |
| "grad_norm": 0.31229549646377563, | |
| "learning_rate": 0.00027052313167259785, | |
| "loss": 0.408, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 3.8917525773195876, | |
| "grad_norm": 0.3603403866291046, | |
| "learning_rate": 0.0002703896797153025, | |
| "loss": 0.3259, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.904639175257732, | |
| "grad_norm": 0.2832438051700592, | |
| "learning_rate": 0.0002702562277580071, | |
| "loss": 0.3423, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 3.917525773195876, | |
| "grad_norm": 0.3435526192188263, | |
| "learning_rate": 0.00027012277580071173, | |
| "loss": 0.2835, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.930412371134021, | |
| "grad_norm": 0.3645428419113159, | |
| "learning_rate": 0.0002699893238434164, | |
| "loss": 0.3483, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 3.943298969072165, | |
| "grad_norm": 0.7062532901763916, | |
| "learning_rate": 0.00026985587188612097, | |
| "loss": 0.2586, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.9561855670103094, | |
| "grad_norm": 0.4606476426124573, | |
| "learning_rate": 0.0002697224199288256, | |
| "loss": 0.3939, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 3.9690721649484537, | |
| "grad_norm": 0.4727219045162201, | |
| "learning_rate": 0.0002695889679715302, | |
| "loss": 0.3151, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.981958762886598, | |
| "grad_norm": 0.2774180471897125, | |
| "learning_rate": 0.00026945551601423486, | |
| "loss": 0.3519, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 3.9948453608247423, | |
| "grad_norm": 0.4793704152107239, | |
| "learning_rate": 0.0002693220640569395, | |
| "loss": 0.3023, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 4.007731958762887, | |
| "grad_norm": 0.3699122965335846, | |
| "learning_rate": 0.0002691886120996441, | |
| "loss": 0.3664, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 4.020618556701031, | |
| "grad_norm": 0.6096347570419312, | |
| "learning_rate": 0.00026905516014234874, | |
| "loss": 0.2579, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 4.033505154639175, | |
| "grad_norm": 0.2851110100746155, | |
| "learning_rate": 0.00026892170818505333, | |
| "loss": 0.3347, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 4.046391752577319, | |
| "grad_norm": 0.5487604141235352, | |
| "learning_rate": 0.000268788256227758, | |
| "loss": 0.2676, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 4.059278350515464, | |
| "grad_norm": 0.29913634061813354, | |
| "learning_rate": 0.0002686548042704626, | |
| "loss": 0.3061, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 4.072164948453608, | |
| "grad_norm": 0.3672349750995636, | |
| "learning_rate": 0.0002685213523131672, | |
| "loss": 0.2632, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 4.085051546391752, | |
| "grad_norm": 0.5969862341880798, | |
| "learning_rate": 0.00026838790035587187, | |
| "loss": 0.2868, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 4.097938144329897, | |
| "grad_norm": 0.4409433901309967, | |
| "learning_rate": 0.0002682544483985765, | |
| "loss": 0.2907, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 4.110824742268041, | |
| "grad_norm": 0.4578838050365448, | |
| "learning_rate": 0.0002681209964412811, | |
| "loss": 0.3377, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 4.123711340206185, | |
| "grad_norm": 0.302491694688797, | |
| "learning_rate": 0.00026798754448398575, | |
| "loss": 0.2857, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.123711340206185, | |
| "eval_loss": 0.4647158980369568, | |
| "eval_runtime": 189.5598, | |
| "eval_samples_per_second": 6.462, | |
| "eval_steps_per_second": 0.812, | |
| "eval_wer": 1.0008163265306123, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.13659793814433, | |
| "grad_norm": 0.5263584852218628, | |
| "learning_rate": 0.00026785409252669034, | |
| "loss": 0.3246, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 4.149484536082475, | |
| "grad_norm": 0.48449578881263733, | |
| "learning_rate": 0.000267720640569395, | |
| "loss": 0.3163, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 4.162371134020619, | |
| "grad_norm": 0.3020855784416199, | |
| "learning_rate": 0.00026758718861209964, | |
| "loss": 0.3248, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 4.175257731958763, | |
| "grad_norm": 0.3554665148258209, | |
| "learning_rate": 0.00026745373665480423, | |
| "loss": 0.2863, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 4.188144329896907, | |
| "grad_norm": 0.4055439531803131, | |
| "learning_rate": 0.0002673202846975089, | |
| "loss": 0.3289, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 4.201030927835052, | |
| "grad_norm": 0.6498619914054871, | |
| "learning_rate": 0.00026718683274021347, | |
| "loss": 0.289, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.213917525773196, | |
| "grad_norm": 0.23331980407238007, | |
| "learning_rate": 0.0002670533807829181, | |
| "loss": 0.2744, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 4.22680412371134, | |
| "grad_norm": 0.5142727494239807, | |
| "learning_rate": 0.00026691992882562276, | |
| "loss": 0.3139, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.239690721649485, | |
| "grad_norm": 0.4468501806259155, | |
| "learning_rate": 0.00026678647686832735, | |
| "loss": 0.3392, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 4.252577319587629, | |
| "grad_norm": 0.36652007699012756, | |
| "learning_rate": 0.000266653024911032, | |
| "loss": 0.2499, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.265463917525773, | |
| "grad_norm": 0.3004089891910553, | |
| "learning_rate": 0.0002665195729537366, | |
| "loss": 0.31, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 4.278350515463917, | |
| "grad_norm": 0.4099670946598053, | |
| "learning_rate": 0.00026638612099644124, | |
| "loss": 0.2798, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.291237113402062, | |
| "grad_norm": 0.26817041635513306, | |
| "learning_rate": 0.0002662526690391459, | |
| "loss": 0.3055, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 4.304123711340206, | |
| "grad_norm": 0.4156699478626251, | |
| "learning_rate": 0.0002661192170818505, | |
| "loss": 0.2942, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.31701030927835, | |
| "grad_norm": 2.611400842666626, | |
| "learning_rate": 0.0002659857651245551, | |
| "loss": 0.3659, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 4.329896907216495, | |
| "grad_norm": 0.46959781646728516, | |
| "learning_rate": 0.00026585231316725977, | |
| "loss": 0.2982, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.342783505154639, | |
| "grad_norm": 0.2867629826068878, | |
| "learning_rate": 0.00026571886120996436, | |
| "loss": 0.3511, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 4.355670103092783, | |
| "grad_norm": 0.38449224829673767, | |
| "learning_rate": 0.000265585409252669, | |
| "loss": 0.2826, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.368556701030927, | |
| "grad_norm": 0.4524473249912262, | |
| "learning_rate": 0.0002654519572953736, | |
| "loss": 0.313, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 4.381443298969073, | |
| "grad_norm": 0.3989889919757843, | |
| "learning_rate": 0.00026531850533807825, | |
| "loss": 0.2825, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.394329896907217, | |
| "grad_norm": 0.3007287085056305, | |
| "learning_rate": 0.0002651850533807829, | |
| "loss": 0.3369, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 4.407216494845361, | |
| "grad_norm": 0.7760800123214722, | |
| "learning_rate": 0.0002650516014234875, | |
| "loss": 0.2835, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.420103092783505, | |
| "grad_norm": 0.27871614694595337, | |
| "learning_rate": 0.00026491814946619214, | |
| "loss": 0.3172, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 4.43298969072165, | |
| "grad_norm": 0.48607179522514343, | |
| "learning_rate": 0.0002647846975088968, | |
| "loss": 0.303, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.445876288659794, | |
| "grad_norm": 0.3540396988391876, | |
| "learning_rate": 0.0002646512455516014, | |
| "loss": 0.355, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 4.458762886597938, | |
| "grad_norm": 1.2004367113113403, | |
| "learning_rate": 0.000264517793594306, | |
| "loss": 0.3197, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 4.471649484536083, | |
| "grad_norm": 0.2405807226896286, | |
| "learning_rate": 0.00026438434163701067, | |
| "loss": 0.3179, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 4.484536082474227, | |
| "grad_norm": 0.36553826928138733, | |
| "learning_rate": 0.00026425088967971526, | |
| "loss": 0.2793, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 4.497422680412371, | |
| "grad_norm": 0.26435115933418274, | |
| "learning_rate": 0.0002641174377224199, | |
| "loss": 0.2849, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 4.510309278350515, | |
| "grad_norm": 0.6196132302284241, | |
| "learning_rate": 0.00026398398576512455, | |
| "loss": 0.3267, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 4.52319587628866, | |
| "grad_norm": 0.2987557649612427, | |
| "learning_rate": 0.00026385053380782915, | |
| "loss": 0.3152, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 4.536082474226804, | |
| "grad_norm": 0.4445992112159729, | |
| "learning_rate": 0.0002637170818505338, | |
| "loss": 0.3, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 4.548969072164948, | |
| "grad_norm": 0.4059930741786957, | |
| "learning_rate": 0.00026358362989323844, | |
| "loss": 0.3395, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 4.561855670103093, | |
| "grad_norm": 0.522637665271759, | |
| "learning_rate": 0.00026345017793594303, | |
| "loss": 0.2783, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 4.574742268041237, | |
| "grad_norm": 0.1963280737400055, | |
| "learning_rate": 0.0002633167259786477, | |
| "loss": 0.2979, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 4.587628865979381, | |
| "grad_norm": 0.7622120976448059, | |
| "learning_rate": 0.00026318327402135227, | |
| "loss": 0.3204, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 4.600515463917525, | |
| "grad_norm": 0.2749217450618744, | |
| "learning_rate": 0.0002630498220640569, | |
| "loss": 0.3365, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 4.61340206185567, | |
| "grad_norm": 0.5664824843406677, | |
| "learning_rate": 0.00026291637010676156, | |
| "loss": 0.2772, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 4.626288659793815, | |
| "grad_norm": 0.3661792278289795, | |
| "learning_rate": 0.00026278291814946616, | |
| "loss": 0.3314, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 4.639175257731958, | |
| "grad_norm": 0.37070122361183167, | |
| "learning_rate": 0.0002626494661921708, | |
| "loss": 0.2991, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.652061855670103, | |
| "grad_norm": 0.4963271915912628, | |
| "learning_rate": 0.00026251601423487545, | |
| "loss": 0.3312, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 4.664948453608248, | |
| "grad_norm": 0.5193173885345459, | |
| "learning_rate": 0.00026238256227758004, | |
| "loss": 0.2914, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 4.677835051546392, | |
| "grad_norm": 0.3158729076385498, | |
| "learning_rate": 0.0002622491103202847, | |
| "loss": 0.3317, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 4.690721649484536, | |
| "grad_norm": 0.37290090322494507, | |
| "learning_rate": 0.00026211565836298934, | |
| "loss": 0.2612, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 4.703608247422681, | |
| "grad_norm": 0.334089070558548, | |
| "learning_rate": 0.00026198220640569393, | |
| "loss": 0.3235, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 4.716494845360825, | |
| "grad_norm": 0.3259222209453583, | |
| "learning_rate": 0.0002618487544483986, | |
| "loss": 0.2952, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 4.729381443298969, | |
| "grad_norm": 0.2776556611061096, | |
| "learning_rate": 0.0002617153024911032, | |
| "loss": 0.3405, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 4.742268041237113, | |
| "grad_norm": 0.33884450793266296, | |
| "learning_rate": 0.0002615818505338078, | |
| "loss": 0.2813, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 4.755154639175258, | |
| "grad_norm": 0.48323071002960205, | |
| "learning_rate": 0.00026144839857651246, | |
| "loss": 0.3044, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 4.768041237113402, | |
| "grad_norm": 0.32996150851249695, | |
| "learning_rate": 0.00026131494661921705, | |
| "loss": 0.2817, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 4.780927835051546, | |
| "grad_norm": 0.3732437491416931, | |
| "learning_rate": 0.0002611814946619217, | |
| "loss": 0.3165, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 4.793814432989691, | |
| "grad_norm": 0.5148000717163086, | |
| "learning_rate": 0.00026104804270462635, | |
| "loss": 0.2636, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 4.806701030927835, | |
| "grad_norm": 0.3256881833076477, | |
| "learning_rate": 0.00026091459074733094, | |
| "loss": 0.3677, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 4.819587628865979, | |
| "grad_norm": 0.5248320698738098, | |
| "learning_rate": 0.0002607811387900356, | |
| "loss": 0.2959, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 4.832474226804123, | |
| "grad_norm": 0.25363728404045105, | |
| "learning_rate": 0.0002606476868327402, | |
| "loss": 0.355, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 4.845360824742268, | |
| "grad_norm": 1.110967755317688, | |
| "learning_rate": 0.0002605142348754448, | |
| "loss": 0.3208, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 4.858247422680412, | |
| "grad_norm": 0.3510916531085968, | |
| "learning_rate": 0.00026038078291814947, | |
| "loss": 0.3421, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 4.871134020618557, | |
| "grad_norm": 0.33312925696372986, | |
| "learning_rate": 0.00026024733096085406, | |
| "loss": 0.3046, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 4.8840206185567006, | |
| "grad_norm": 0.32453814148902893, | |
| "learning_rate": 0.0002601138790035587, | |
| "loss": 0.3294, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 4.896907216494846, | |
| "grad_norm": 0.3632301092147827, | |
| "learning_rate": 0.0002599804270462633, | |
| "loss": 0.3261, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.90979381443299, | |
| "grad_norm": 0.3007452189922333, | |
| "learning_rate": 0.00025984697508896795, | |
| "loss": 0.3524, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 4.922680412371134, | |
| "grad_norm": 0.47182586789131165, | |
| "learning_rate": 0.0002597135231316726, | |
| "loss": 0.3287, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 4.935567010309279, | |
| "grad_norm": 0.3959861397743225, | |
| "learning_rate": 0.0002595800711743772, | |
| "loss": 0.3134, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 4.948453608247423, | |
| "grad_norm": 0.28187814354896545, | |
| "learning_rate": 0.00025944661921708183, | |
| "loss": 0.3154, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.961340206185567, | |
| "grad_norm": 0.32686540484428406, | |
| "learning_rate": 0.0002593131672597865, | |
| "loss": 0.3464, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 4.974226804123711, | |
| "grad_norm": 0.46848079562187195, | |
| "learning_rate": 0.0002591797153024911, | |
| "loss": 0.2753, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 4.987113402061856, | |
| "grad_norm": 0.25264793634414673, | |
| "learning_rate": 0.0002590462633451957, | |
| "loss": 0.3403, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.4163696765899658, | |
| "learning_rate": 0.0002589128113879003, | |
| "loss": 0.2669, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 5.012886597938144, | |
| "grad_norm": 0.29742294549942017, | |
| "learning_rate": 0.00025877935943060496, | |
| "loss": 0.3203, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 5.025773195876289, | |
| "grad_norm": 0.8651419281959534, | |
| "learning_rate": 0.0002586459074733096, | |
| "loss": 0.2589, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 5.038659793814433, | |
| "grad_norm": 0.2916441261768341, | |
| "learning_rate": 0.0002585124555160142, | |
| "loss": 0.3, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 5.051546391752577, | |
| "grad_norm": 1.4987293481826782, | |
| "learning_rate": 0.00025837900355871884, | |
| "loss": 0.3148, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 5.064432989690721, | |
| "grad_norm": 0.46689650416374207, | |
| "learning_rate": 0.00025824555160142344, | |
| "loss": 0.2935, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 5.077319587628866, | |
| "grad_norm": 0.3726617991924286, | |
| "learning_rate": 0.0002581120996441281, | |
| "loss": 0.2871, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 5.09020618556701, | |
| "grad_norm": 0.29355180263519287, | |
| "learning_rate": 0.00025797864768683273, | |
| "loss": 0.2422, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 5.103092783505154, | |
| "grad_norm": 0.6249547600746155, | |
| "learning_rate": 0.0002578451957295373, | |
| "loss": 0.2749, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 5.115979381443299, | |
| "grad_norm": 0.42613521218299866, | |
| "learning_rate": 0.00025771174377224197, | |
| "loss": 0.2998, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 5.128865979381443, | |
| "grad_norm": 0.4285449981689453, | |
| "learning_rate": 0.00025757829181494656, | |
| "loss": 0.3241, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 5.141752577319588, | |
| "grad_norm": 0.2756471037864685, | |
| "learning_rate": 0.0002574448398576512, | |
| "loss": 0.2788, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 5.154639175257732, | |
| "grad_norm": 0.650787889957428, | |
| "learning_rate": 0.00025731138790035586, | |
| "loss": 0.2628, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.154639175257732, | |
| "eval_loss": 0.5844400525093079, | |
| "eval_runtime": 196.2011, | |
| "eval_samples_per_second": 6.244, | |
| "eval_steps_per_second": 0.785, | |
| "eval_wer": 1.006530612244898, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.154639175257732, | |
| "step": 10000, | |
| "total_flos": 5.52048529474528e+19, | |
| "train_loss": 0.6429409675121307, | |
| "train_runtime": 62307.8234, | |
| "train_samples_per_second": 22.413, | |
| "train_steps_per_second": 0.934 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 58200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 30, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.52048529474528e+19, | |
| "train_batch_size": 12, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |