| { | |
| "best_metric": 0.04373383894562721, | |
| "best_model_checkpoint": "w2v-bert-final-v2/checkpoint-10000", | |
| "epoch": 7.716049382716049, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.23148148148148148, | |
| "grad_norm": 7.229776859283447, | |
| "learning_rate": 7.425e-06, | |
| "loss": 3.4047, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.46296296296296297, | |
| "grad_norm": 11.000567436218262, | |
| "learning_rate": 1.4925e-05, | |
| "loss": 0.8648, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6944444444444444, | |
| "grad_norm": 11.643492698669434, | |
| "learning_rate": 2.2400000000000002e-05, | |
| "loss": 0.6263, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7716049382716049, | |
| "eval_loss": 0.5938597321510315, | |
| "eval_runtime": 49.2872, | |
| "eval_samples_per_second": 38.408, | |
| "eval_steps_per_second": 4.809, | |
| "eval_wer": 0.509299543180335, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.9259259259259259, | |
| "grad_norm": 16.209674835205078, | |
| "learning_rate": 2.9900000000000002e-05, | |
| "loss": 0.5512, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.1574074074074074, | |
| "grad_norm": 2.7776882648468018, | |
| "learning_rate": 3.74e-05, | |
| "loss": 0.4844, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 5.179060459136963, | |
| "learning_rate": 4.4875e-05, | |
| "loss": 0.4641, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.5432098765432098, | |
| "eval_loss": 0.5058629512786865, | |
| "eval_runtime": 49.9367, | |
| "eval_samples_per_second": 37.908, | |
| "eval_steps_per_second": 4.746, | |
| "eval_wer": 0.45061996954535566, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.6203703703703702, | |
| "grad_norm": 2.7534990310668945, | |
| "learning_rate": 4.943236137667304e-05, | |
| "loss": 0.4629, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.8518518518518519, | |
| "grad_norm": 3.1153225898742676, | |
| "learning_rate": 4.763981835564054e-05, | |
| "loss": 0.4393, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.0833333333333335, | |
| "grad_norm": 2.560495376586914, | |
| "learning_rate": 4.584727533460803e-05, | |
| "loss": 0.3782, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.314814814814815, | |
| "grad_norm": 2.7176785469055176, | |
| "learning_rate": 4.4054732313575525e-05, | |
| "loss": 0.3054, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.314814814814815, | |
| "eval_loss": 0.34695935249328613, | |
| "eval_runtime": 49.4937, | |
| "eval_samples_per_second": 38.247, | |
| "eval_steps_per_second": 4.788, | |
| "eval_wer": 0.3308135740700457, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.5462962962962963, | |
| "grad_norm": 2.319594383239746, | |
| "learning_rate": 4.226816443594647e-05, | |
| "loss": 0.3079, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 2.165224552154541, | |
| "learning_rate": 4.047562141491396e-05, | |
| "loss": 0.301, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.009259259259259, | |
| "grad_norm": 3.609168529510498, | |
| "learning_rate": 3.8683078393881456e-05, | |
| "loss": 0.2837, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.0864197530864197, | |
| "eval_loss": 0.2659013867378235, | |
| "eval_runtime": 50.1986, | |
| "eval_samples_per_second": 37.71, | |
| "eval_steps_per_second": 4.721, | |
| "eval_wer": 0.2583206438981945, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.240740740740741, | |
| "grad_norm": 1.6436126232147217, | |
| "learning_rate": 3.689053537284895e-05, | |
| "loss": 0.2191, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.4722222222222223, | |
| "grad_norm": 1.704137921333313, | |
| "learning_rate": 3.5103967495219884e-05, | |
| "loss": 0.2103, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.7037037037037037, | |
| "grad_norm": 6.373330593109131, | |
| "learning_rate": 3.331142447418738e-05, | |
| "loss": 0.2174, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.8580246913580245, | |
| "eval_loss": 0.19009661674499512, | |
| "eval_runtime": 49.6613, | |
| "eval_samples_per_second": 38.118, | |
| "eval_steps_per_second": 4.772, | |
| "eval_wer": 0.195072873613226, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.935185185185185, | |
| "grad_norm": 2.4362969398498535, | |
| "learning_rate": 3.151888145315488e-05, | |
| "loss": 0.1939, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 4.166666666666667, | |
| "grad_norm": 1.4299020767211914, | |
| "learning_rate": 2.9726338432122373e-05, | |
| "loss": 0.1666, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 4.398148148148148, | |
| "grad_norm": 1.3548126220703125, | |
| "learning_rate": 2.793379541108987e-05, | |
| "loss": 0.1569, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 4.62962962962963, | |
| "grad_norm": 1.5247106552124023, | |
| "learning_rate": 2.6141252390057363e-05, | |
| "loss": 0.152, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.62962962962963, | |
| "eval_loss": 0.15165844559669495, | |
| "eval_runtime": 49.8658, | |
| "eval_samples_per_second": 37.962, | |
| "eval_steps_per_second": 4.753, | |
| "eval_wer": 0.1540678703502284, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.861111111111111, | |
| "grad_norm": 3.228041410446167, | |
| "learning_rate": 2.434870936902486e-05, | |
| "loss": 0.1481, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 5.092592592592593, | |
| "grad_norm": 1.5059906244277954, | |
| "learning_rate": 2.2562141491395794e-05, | |
| "loss": 0.1256, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 5.324074074074074, | |
| "grad_norm": 3.538102626800537, | |
| "learning_rate": 2.0769598470363287e-05, | |
| "loss": 0.1089, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 5.401234567901234, | |
| "eval_loss": 0.11081259697675705, | |
| "eval_runtime": 49.8031, | |
| "eval_samples_per_second": 38.01, | |
| "eval_steps_per_second": 4.759, | |
| "eval_wer": 0.11404176636937133, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 0.6531468033790588, | |
| "learning_rate": 1.8977055449330787e-05, | |
| "loss": 0.1022, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 5.787037037037037, | |
| "grad_norm": 1.7070139646530151, | |
| "learning_rate": 1.718451242829828e-05, | |
| "loss": 0.1022, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 6.018518518518518, | |
| "grad_norm": 1.612425446510315, | |
| "learning_rate": 1.5397944550669215e-05, | |
| "loss": 0.0986, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 6.172839506172839, | |
| "eval_loss": 0.08366883546113968, | |
| "eval_runtime": 51.7773, | |
| "eval_samples_per_second": 36.56, | |
| "eval_steps_per_second": 4.577, | |
| "eval_wer": 0.08902545138133565, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "grad_norm": 0.7614215612411499, | |
| "learning_rate": 1.3605401529636713e-05, | |
| "loss": 0.0766, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 6.481481481481482, | |
| "grad_norm": 2.542365550994873, | |
| "learning_rate": 1.181883365200765e-05, | |
| "loss": 0.0758, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 6.712962962962963, | |
| "grad_norm": 1.319675326347351, | |
| "learning_rate": 1.0026290630975144e-05, | |
| "loss": 0.0726, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 6.944444444444445, | |
| "grad_norm": 0.8109455704689026, | |
| "learning_rate": 8.233747609942639e-06, | |
| "loss": 0.0648, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 6.944444444444445, | |
| "eval_loss": 0.05808680132031441, | |
| "eval_runtime": 51.2984, | |
| "eval_samples_per_second": 36.902, | |
| "eval_steps_per_second": 4.62, | |
| "eval_wer": 0.06031107243854688, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 7.175925925925926, | |
| "grad_norm": 1.2560392618179321, | |
| "learning_rate": 6.441204588910134e-06, | |
| "loss": 0.051, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 7.407407407407407, | |
| "grad_norm": 3.7403156757354736, | |
| "learning_rate": 4.654636711281071e-06, | |
| "loss": 0.0473, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 7.638888888888889, | |
| "grad_norm": 0.7289965748786926, | |
| "learning_rate": 2.862093690248566e-06, | |
| "loss": 0.0499, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 7.716049382716049, | |
| "eval_loss": 0.04373383894562721, | |
| "eval_runtime": 50.2864, | |
| "eval_samples_per_second": 37.644, | |
| "eval_steps_per_second": 4.713, | |
| "eval_wer": 0.04606264955405699, | |
| "step": 10000 | |
| } | |
| ], | |
| "logging_steps": 300, | |
| "max_steps": 10368, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 8, | |
| "save_steps": 1000, | |
| "total_flos": 2.075992650724199e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |