| { | |
| "best_metric": 0.9677416086196899, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.014399107255350168, | |
| "eval_steps": 25, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00028798214510700334, | |
| "grad_norm": 0.4370923638343811, | |
| "learning_rate": 5e-05, | |
| "loss": 0.8523, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00028798214510700334, | |
| "eval_loss": 1.954984188079834, | |
| "eval_runtime": 1331.5904, | |
| "eval_samples_per_second": 17.568, | |
| "eval_steps_per_second": 2.197, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005759642902140067, | |
| "grad_norm": 0.42163851857185364, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9454, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0008639464353210101, | |
| "grad_norm": 0.48534369468688965, | |
| "learning_rate": 9.989294616193017e-05, | |
| "loss": 0.9911, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0011519285804280134, | |
| "grad_norm": 0.38114985823631287, | |
| "learning_rate": 9.957224306869053e-05, | |
| "loss": 1.0287, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0014399107255350168, | |
| "grad_norm": 0.4735317826271057, | |
| "learning_rate": 9.903926402016153e-05, | |
| "loss": 1.1522, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0017278928706420203, | |
| "grad_norm": 0.5327126383781433, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 1.1903, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0020158750157490235, | |
| "grad_norm": 0.4486357867717743, | |
| "learning_rate": 9.73465064747553e-05, | |
| "loss": 1.1209, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0023038571608560267, | |
| "grad_norm": 0.4383220672607422, | |
| "learning_rate": 9.619397662556435e-05, | |
| "loss": 1.156, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0025918393059630304, | |
| "grad_norm": 0.530921459197998, | |
| "learning_rate": 9.484363707663442e-05, | |
| "loss": 1.0935, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0028798214510700336, | |
| "grad_norm": 0.5392281413078308, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 1.1428, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.003167803596177037, | |
| "grad_norm": 0.7135190367698669, | |
| "learning_rate": 9.157348061512727e-05, | |
| "loss": 1.1796, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0034557857412840405, | |
| "grad_norm": 0.7297762036323547, | |
| "learning_rate": 8.966766701456177e-05, | |
| "loss": 1.1658, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0037437678863910437, | |
| "grad_norm": 0.5865046977996826, | |
| "learning_rate": 8.759199037394887e-05, | |
| "loss": 0.6497, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.004031750031498047, | |
| "grad_norm": 0.24555620551109314, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 0.6587, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.004319732176605051, | |
| "grad_norm": 0.3134182095527649, | |
| "learning_rate": 8.296729075500344e-05, | |
| "loss": 0.7629, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.004607714321712053, | |
| "grad_norm": 0.36211544275283813, | |
| "learning_rate": 8.043807145043604e-05, | |
| "loss": 0.814, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.004895696466819057, | |
| "grad_norm": 0.39566537737846375, | |
| "learning_rate": 7.777851165098012e-05, | |
| "loss": 0.8591, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.005183678611926061, | |
| "grad_norm": 0.301882266998291, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.8022, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0054716607570330636, | |
| "grad_norm": 0.2724820375442505, | |
| "learning_rate": 7.211443451095007e-05, | |
| "loss": 0.8338, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.005759642902140067, | |
| "grad_norm": 0.27424779534339905, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 0.9515, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.006047625047247071, | |
| "grad_norm": 0.2924113869667053, | |
| "learning_rate": 6.607197326515808e-05, | |
| "loss": 0.9032, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.006335607192354074, | |
| "grad_norm": 0.32882699370384216, | |
| "learning_rate": 6.294095225512603e-05, | |
| "loss": 0.9395, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.006623589337461077, | |
| "grad_norm": 0.3634035885334015, | |
| "learning_rate": 5.9754516100806423e-05, | |
| "loss": 1.0316, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.006911571482568081, | |
| "grad_norm": 0.3925745189189911, | |
| "learning_rate": 5.6526309611002594e-05, | |
| "loss": 1.0829, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.007199553627675084, | |
| "grad_norm": 0.44211986660957336, | |
| "learning_rate": 5.327015646150716e-05, | |
| "loss": 1.0893, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.007199553627675084, | |
| "eval_loss": 0.9996042251586914, | |
| "eval_runtime": 1325.9628, | |
| "eval_samples_per_second": 17.642, | |
| "eval_steps_per_second": 2.206, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0074875357727820875, | |
| "grad_norm": 0.11079126596450806, | |
| "learning_rate": 5e-05, | |
| "loss": 0.6212, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.007775517917889091, | |
| "grad_norm": 0.13816042244434357, | |
| "learning_rate": 4.6729843538492847e-05, | |
| "loss": 0.7489, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.008063500062996094, | |
| "grad_norm": 0.1543881893157959, | |
| "learning_rate": 4.347369038899744e-05, | |
| "loss": 0.792, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.008351482208103098, | |
| "grad_norm": 0.18760843575000763, | |
| "learning_rate": 4.0245483899193595e-05, | |
| "loss": 0.7768, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.008639464353210101, | |
| "grad_norm": 0.21165847778320312, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 0.7727, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.008927446498317105, | |
| "grad_norm": 0.21181799471378326, | |
| "learning_rate": 3.392802673484193e-05, | |
| "loss": 0.8389, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.009215428643424107, | |
| "grad_norm": 0.19947344064712524, | |
| "learning_rate": 3.086582838174551e-05, | |
| "loss": 0.859, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.00950341078853111, | |
| "grad_norm": 0.2230692207813263, | |
| "learning_rate": 2.7885565489049946e-05, | |
| "loss": 0.8318, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.009791392933638114, | |
| "grad_norm": 0.2214939445257187, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 0.8046, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.010079375078745118, | |
| "grad_norm": 0.2411639243364334, | |
| "learning_rate": 2.2221488349019903e-05, | |
| "loss": 1.0056, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.010367357223852122, | |
| "grad_norm": 0.313069611787796, | |
| "learning_rate": 1.9561928549563968e-05, | |
| "loss": 1.1206, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.010655339368959125, | |
| "grad_norm": 0.4024619460105896, | |
| "learning_rate": 1.703270924499656e-05, | |
| "loss": 1.1273, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.010943321514066127, | |
| "grad_norm": 0.09190559387207031, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 0.6195, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.01123130365917313, | |
| "grad_norm": 0.11642751097679138, | |
| "learning_rate": 1.2408009626051137e-05, | |
| "loss": 0.6657, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.011519285804280134, | |
| "grad_norm": 0.11892105638980865, | |
| "learning_rate": 1.0332332985438248e-05, | |
| "loss": 0.6754, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.011807267949387138, | |
| "grad_norm": 0.1529056876897812, | |
| "learning_rate": 8.426519384872733e-06, | |
| "loss": 0.8458, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.012095250094494142, | |
| "grad_norm": 0.14376507699489594, | |
| "learning_rate": 6.698729810778065e-06, | |
| "loss": 0.7614, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.012383232239601145, | |
| "grad_norm": 0.16135233640670776, | |
| "learning_rate": 5.156362923365588e-06, | |
| "loss": 0.7901, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.012671214384708147, | |
| "grad_norm": 0.1866428554058075, | |
| "learning_rate": 3.8060233744356633e-06, | |
| "loss": 0.7842, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.012959196529815151, | |
| "grad_norm": 0.18563878536224365, | |
| "learning_rate": 2.653493525244721e-06, | |
| "loss": 0.8421, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.013247178674922155, | |
| "grad_norm": 0.25013992190361023, | |
| "learning_rate": 1.70370868554659e-06, | |
| "loss": 0.8851, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.013535160820029158, | |
| "grad_norm": 0.24550536274909973, | |
| "learning_rate": 9.607359798384785e-07, | |
| "loss": 0.8933, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.013823142965136162, | |
| "grad_norm": 0.25731024146080017, | |
| "learning_rate": 4.277569313094809e-07, | |
| "loss": 0.9443, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.014111125110243166, | |
| "grad_norm": 0.31453737616539, | |
| "learning_rate": 1.0705383806982606e-07, | |
| "loss": 0.99, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.014399107255350168, | |
| "grad_norm": 0.3844415545463562, | |
| "learning_rate": 0.0, | |
| "loss": 1.1618, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.014399107255350168, | |
| "eval_loss": 0.9677416086196899, | |
| "eval_runtime": 1330.2701, | |
| "eval_samples_per_second": 17.585, | |
| "eval_steps_per_second": 2.199, | |
| "step": 50 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 50, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 1, | |
| "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": 2.6947910513564058e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |