| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 49.95402298850575, | |
| "global_step": 8692, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.123417721518986e-06, | |
| "loss": 14.6554, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.766772151898734e-05, | |
| "loss": 6.292, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.6212025316455694e-05, | |
| "loss": 4.5602, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.475632911392405e-05, | |
| "loss": 3.675, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "eval_cer": 1.0, | |
| "eval_loss": 3.505185604095459, | |
| "eval_runtime": 388.8147, | |
| "eval_samples_per_second": 19.927, | |
| "eval_steps_per_second": 0.314, | |
| "eval_wer": 1.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.3300632911392406e-05, | |
| "loss": 3.4053, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 5.184493670886076e-05, | |
| "loss": 3.3269, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.038924050632911e-05, | |
| "loss": 3.2702, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 6.893354430379747e-05, | |
| "loss": 3.0446, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "eval_cer": 0.52146416703672, | |
| "eval_loss": 2.2759008407592773, | |
| "eval_runtime": 391.3771, | |
| "eval_samples_per_second": 19.797, | |
| "eval_steps_per_second": 0.312, | |
| "eval_wer": 1.0052476497825171, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 7.5e-05, | |
| "loss": 2.3045, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.9583, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.8142, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.7276, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "eval_cer": 0.19686409281152592, | |
| "eval_loss": 0.7083391547203064, | |
| "eval_runtime": 386.6895, | |
| "eval_samples_per_second": 20.037, | |
| "eval_steps_per_second": 0.315, | |
| "eval_wer": 0.6697067489827417, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 7.47, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.6571, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.6004, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.5572, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.5171, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "eval_cer": 0.15683826416017196, | |
| "eval_loss": 0.5328246355056763, | |
| "eval_runtime": 388.9054, | |
| "eval_samples_per_second": 19.923, | |
| "eval_steps_per_second": 0.314, | |
| "eval_wer": 0.5733408166128806, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.4951, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.4553, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.4348, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.4176, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "eval_cer": 0.13810176574442937, | |
| "eval_loss": 0.4571371376514435, | |
| "eval_runtime": 390.2121, | |
| "eval_samples_per_second": 19.856, | |
| "eval_steps_per_second": 0.313, | |
| "eval_wer": 0.5160656657780273, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.3988, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 12.64, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.3666, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 13.22, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.3585, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.343, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "eval_cer": 0.11596506696356744, | |
| "eval_loss": 0.3910418748855591, | |
| "eval_runtime": 386.6143, | |
| "eval_samples_per_second": 20.041, | |
| "eval_steps_per_second": 0.316, | |
| "eval_wer": 0.45223796828960294, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.3196, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.3085, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 15.52, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.3006, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 16.09, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2743, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 16.09, | |
| "eval_cer": 0.10436362622251684, | |
| "eval_loss": 0.3533952534198761, | |
| "eval_runtime": 386.8689, | |
| "eval_samples_per_second": 20.027, | |
| "eval_steps_per_second": 0.315, | |
| "eval_wer": 0.41365230812403536, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.256, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 17.24, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2627, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 17.82, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2496, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 18.39, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2396, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 18.39, | |
| "eval_cer": 0.09588994077001131, | |
| "eval_loss": 0.3277980387210846, | |
| "eval_runtime": 388.8491, | |
| "eval_samples_per_second": 19.925, | |
| "eval_steps_per_second": 0.314, | |
| "eval_wer": 0.38773677564192505, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2291, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 19.54, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2254, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 20.11, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2063, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 20.69, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.2035, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 20.69, | |
| "eval_cer": 0.09172603609575479, | |
| "eval_loss": 0.31091800332069397, | |
| "eval_runtime": 384.3384, | |
| "eval_samples_per_second": 20.159, | |
| "eval_steps_per_second": 0.317, | |
| "eval_wer": 0.3740844675178897, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 21.26, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1937, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 21.84, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1787, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 22.41, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1779, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 22.99, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1745, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 22.99, | |
| "eval_cer": 0.08818425010028982, | |
| "eval_loss": 0.2971595227718353, | |
| "eval_runtime": 385.6559, | |
| "eval_samples_per_second": 20.09, | |
| "eval_steps_per_second": 0.316, | |
| "eval_wer": 0.3618212431598148, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 23.56, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1697, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 24.14, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1506, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 24.71, | |
| "learning_rate": 7.5e-05, | |
| "loss": 1.1512, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 25.29, | |
| "learning_rate": 7.414768806073154e-05, | |
| "loss": 1.1541, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 25.29, | |
| "eval_cer": 0.08323089827501506, | |
| "eval_loss": 0.28362834453582764, | |
| "eval_runtime": 385.7883, | |
| "eval_samples_per_second": 20.084, | |
| "eval_steps_per_second": 0.316, | |
| "eval_wer": 0.3427108180159955, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 25.86, | |
| "learning_rate": 7.250862663906142e-05, | |
| "loss": 1.1464, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 26.44, | |
| "learning_rate": 7.08695652173913e-05, | |
| "loss": 1.1395, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 6.923050379572118e-05, | |
| "loss": 1.1432, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 27.59, | |
| "learning_rate": 6.759144237405107e-05, | |
| "loss": 1.1372, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 27.59, | |
| "eval_cer": 0.08124869945017581, | |
| "eval_loss": 0.2759494483470917, | |
| "eval_runtime": 383.6861, | |
| "eval_samples_per_second": 20.194, | |
| "eval_steps_per_second": 0.318, | |
| "eval_wer": 0.3357373368878911, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 28.16, | |
| "learning_rate": 6.595238095238095e-05, | |
| "loss": 1.1179, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 28.74, | |
| "learning_rate": 6.431331953071083e-05, | |
| "loss": 1.1221, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 29.31, | |
| "learning_rate": 6.267425810904071e-05, | |
| "loss": 1.116, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 29.89, | |
| "learning_rate": 6.103519668737059e-05, | |
| "loss": 1.1048, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 29.89, | |
| "eval_cer": 0.07828827263385742, | |
| "eval_loss": 0.2669001519680023, | |
| "eval_runtime": 386.0726, | |
| "eval_samples_per_second": 20.069, | |
| "eval_steps_per_second": 0.316, | |
| "eval_wer": 0.32842710818015997, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 30.46, | |
| "learning_rate": 5.939613526570048e-05, | |
| "loss": 1.1005, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 31.03, | |
| "learning_rate": 5.775707384403037e-05, | |
| "loss": 1.1056, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 31.61, | |
| "learning_rate": 5.6118012422360246e-05, | |
| "loss": 1.093, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 32.18, | |
| "learning_rate": 5.4478951000690126e-05, | |
| "loss": 1.0966, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 32.18, | |
| "eval_cer": 0.07754173021930756, | |
| "eval_loss": 0.2677817642688751, | |
| "eval_runtime": 386.0716, | |
| "eval_samples_per_second": 20.069, | |
| "eval_steps_per_second": 0.316, | |
| "eval_wer": 0.32491932089238107, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 32.76, | |
| "learning_rate": 5.283988957902001e-05, | |
| "loss": 1.0884, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 33.33, | |
| "learning_rate": 5.120082815734989e-05, | |
| "loss": 1.0878, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 33.91, | |
| "learning_rate": 4.956176673567977e-05, | |
| "loss": 1.0803, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 34.48, | |
| "learning_rate": 4.792270531400966e-05, | |
| "loss": 1.0747, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 34.48, | |
| "eval_cer": 0.07477866519074373, | |
| "eval_loss": 0.25474071502685547, | |
| "eval_runtime": 385.0232, | |
| "eval_samples_per_second": 20.123, | |
| "eval_steps_per_second": 0.317, | |
| "eval_wer": 0.31337168514101305, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 35.06, | |
| "learning_rate": 4.628364389233954e-05, | |
| "loss": 1.0704, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 35.63, | |
| "learning_rate": 4.464458247066942e-05, | |
| "loss": 1.0707, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 36.21, | |
| "learning_rate": 4.30055210489993e-05, | |
| "loss": 1.0599, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 36.78, | |
| "learning_rate": 4.136645962732919e-05, | |
| "loss": 1.0593, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 36.78, | |
| "eval_cer": 0.07282435444460891, | |
| "eval_loss": 0.2490725815296173, | |
| "eval_runtime": 384.9593, | |
| "eval_samples_per_second": 20.127, | |
| "eval_steps_per_second": 0.317, | |
| "eval_wer": 0.3077311631822646, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 37.36, | |
| "learning_rate": 3.972739820565908e-05, | |
| "loss": 1.0572, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 37.93, | |
| "learning_rate": 3.808833678398896e-05, | |
| "loss": 1.0506, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 38.51, | |
| "learning_rate": 3.644927536231883e-05, | |
| "loss": 1.0478, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 39.08, | |
| "learning_rate": 3.4810213940648726e-05, | |
| "loss": 1.0417, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 39.08, | |
| "eval_cer": 0.07105667930211156, | |
| "eval_loss": 0.24495387077331543, | |
| "eval_runtime": 382.2473, | |
| "eval_samples_per_second": 20.27, | |
| "eval_steps_per_second": 0.319, | |
| "eval_wer": 0.301220709976147, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 39.66, | |
| "learning_rate": 3.3171152518978605e-05, | |
| "loss": 1.0359, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 40.23, | |
| "learning_rate": 3.153209109730849e-05, | |
| "loss": 1.0389, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 40.8, | |
| "learning_rate": 2.989302967563837e-05, | |
| "loss": 1.0365, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 41.38, | |
| "learning_rate": 2.8253968253968247e-05, | |
| "loss": 1.024, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 41.38, | |
| "eval_cer": 0.06938768505349148, | |
| "eval_loss": 0.24019765853881836, | |
| "eval_runtime": 381.671, | |
| "eval_samples_per_second": 20.3, | |
| "eval_steps_per_second": 0.32, | |
| "eval_wer": 0.29560825031570087, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 41.95, | |
| "learning_rate": 2.6614906832298136e-05, | |
| "loss": 1.0288, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 42.53, | |
| "learning_rate": 2.4975845410628016e-05, | |
| "loss": 1.0283, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 43.1, | |
| "learning_rate": 2.3353174603174605e-05, | |
| "loss": 1.0153, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 43.68, | |
| "learning_rate": 2.1714113181504485e-05, | |
| "loss": 1.0106, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 43.68, | |
| "eval_cer": 0.06811555961720395, | |
| "eval_loss": 0.23507660627365112, | |
| "eval_runtime": 383.1253, | |
| "eval_samples_per_second": 20.223, | |
| "eval_steps_per_second": 0.318, | |
| "eval_wer": 0.29151115476357514, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 44.25, | |
| "learning_rate": 2.007505175983437e-05, | |
| "loss": 1.013, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 44.83, | |
| "learning_rate": 1.843599033816425e-05, | |
| "loss": 1.0067, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 45.4, | |
| "learning_rate": 1.6796928916494137e-05, | |
| "loss": 1.0076, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 45.98, | |
| "learning_rate": 1.5157867494824016e-05, | |
| "loss": 1.0014, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 45.98, | |
| "eval_cer": 0.06731109580842178, | |
| "eval_loss": 0.23282095789909363, | |
| "eval_runtime": 382.9081, | |
| "eval_samples_per_second": 20.235, | |
| "eval_steps_per_second": 0.319, | |
| "eval_wer": 0.2896309807773257, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 46.55, | |
| "learning_rate": 1.3518806073153895e-05, | |
| "loss": 0.9955, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 47.13, | |
| "learning_rate": 1.1879744651483783e-05, | |
| "loss": 0.9998, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 47.7, | |
| "learning_rate": 1.0240683229813662e-05, | |
| "loss": 0.9956, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 48.28, | |
| "learning_rate": 8.60162180814355e-06, | |
| "loss": 0.9999, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 48.28, | |
| "eval_cer": 0.06668254141915997, | |
| "eval_loss": 0.2318294197320938, | |
| "eval_runtime": 383.814, | |
| "eval_samples_per_second": 20.187, | |
| "eval_steps_per_second": 0.318, | |
| "eval_wer": 0.2865862214115336, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 48.85, | |
| "learning_rate": 6.962560386473429e-06, | |
| "loss": 0.9898, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 49.43, | |
| "learning_rate": 5.323498964803316e-06, | |
| "loss": 0.997, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 49.95, | |
| "step": 8692, | |
| "total_flos": 2.038617513286856e+20, | |
| "train_loss": 1.5808194972818148, | |
| "train_runtime": 72373.5093, | |
| "train_samples_per_second": 15.373, | |
| "train_steps_per_second": 0.12 | |
| } | |
| ], | |
| "max_steps": 8692, | |
| "num_train_epochs": 50, | |
| "total_flos": 2.038617513286856e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |