| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.23346303501945526, |
| "eval_steps": 500, |
| "global_step": 60, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0038910505836575876, |
| "grad_norm": 1.7394803762435913, |
| "learning_rate": 4e-05, |
| "loss": 3.7559, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.007782101167315175, |
| "grad_norm": 1.8431400060653687, |
| "learning_rate": 8e-05, |
| "loss": 3.9259, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.011673151750972763, |
| "grad_norm": 1.673993468284607, |
| "learning_rate": 0.00012, |
| "loss": 3.5901, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.01556420233463035, |
| "grad_norm": 1.7249231338500977, |
| "learning_rate": 0.00016, |
| "loss": 3.3902, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.019455252918287938, |
| "grad_norm": 1.7279131412506104, |
| "learning_rate": 0.0002, |
| "loss": 3.159, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.023346303501945526, |
| "grad_norm": 1.971978783607483, |
| "learning_rate": 0.00019636363636363636, |
| "loss": 2.4741, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.027237354085603113, |
| "grad_norm": 2.6057095527648926, |
| "learning_rate": 0.00019272727272727274, |
| "loss": 2.0208, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.0311284046692607, |
| "grad_norm": 2.154906988143921, |
| "learning_rate": 0.0001890909090909091, |
| "loss": 1.5799, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03501945525291829, |
| "grad_norm": 1.883897304534912, |
| "learning_rate": 0.00018545454545454545, |
| "loss": 1.3594, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.038910505836575876, |
| "grad_norm": 1.5267574787139893, |
| "learning_rate": 0.00018181818181818183, |
| "loss": 0.9735, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.042801556420233464, |
| "grad_norm": 1.3312914371490479, |
| "learning_rate": 0.0001781818181818182, |
| "loss": 1.0511, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.04669260700389105, |
| "grad_norm": 0.8513796329498291, |
| "learning_rate": 0.00017454545454545454, |
| "loss": 0.9663, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.05058365758754864, |
| "grad_norm": 1.2995551824569702, |
| "learning_rate": 0.0001709090909090909, |
| "loss": 0.824, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.054474708171206226, |
| "grad_norm": 0.9970191717147827, |
| "learning_rate": 0.00016727272727272728, |
| "loss": 0.8052, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.058365758754863814, |
| "grad_norm": 1.361310362815857, |
| "learning_rate": 0.00016363636363636366, |
| "loss": 0.887, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.0622568093385214, |
| "grad_norm": 1.3233678340911865, |
| "learning_rate": 0.00016, |
| "loss": 0.789, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06614785992217899, |
| "grad_norm": 0.8609960079193115, |
| "learning_rate": 0.00015636363636363637, |
| "loss": 0.7533, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.07003891050583658, |
| "grad_norm": 0.8982454538345337, |
| "learning_rate": 0.00015272727272727275, |
| "loss": 0.7357, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.07392996108949416, |
| "grad_norm": 0.8766241669654846, |
| "learning_rate": 0.0001490909090909091, |
| "loss": 0.7128, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.07782101167315175, |
| "grad_norm": 0.9981120228767395, |
| "learning_rate": 0.00014545454545454546, |
| "loss": 0.7562, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08171206225680934, |
| "grad_norm": 0.8989161252975464, |
| "learning_rate": 0.00014181818181818184, |
| "loss": 0.6923, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.08560311284046693, |
| "grad_norm": 0.9787339568138123, |
| "learning_rate": 0.0001381818181818182, |
| "loss": 0.7323, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.08949416342412451, |
| "grad_norm": 1.0249125957489014, |
| "learning_rate": 0.00013454545454545455, |
| "loss": 0.7518, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.0933852140077821, |
| "grad_norm": 1.141008734703064, |
| "learning_rate": 0.00013090909090909093, |
| "loss": 0.5975, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.09727626459143969, |
| "grad_norm": 1.1703013181686401, |
| "learning_rate": 0.00012727272727272728, |
| "loss": 0.6332, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.10116731517509728, |
| "grad_norm": 1.2739671468734741, |
| "learning_rate": 0.00012363636363636364, |
| "loss": 0.633, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10505836575875487, |
| "grad_norm": 1.3738476037979126, |
| "learning_rate": 0.00012, |
| "loss": 0.704, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.10894941634241245, |
| "grad_norm": 1.2752115726470947, |
| "learning_rate": 0.00011636363636363636, |
| "loss": 0.5509, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11284046692607004, |
| "grad_norm": 1.367374300956726, |
| "learning_rate": 0.00011272727272727272, |
| "loss": 0.6592, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.11673151750972763, |
| "grad_norm": 1.1849249601364136, |
| "learning_rate": 0.00010909090909090909, |
| "loss": 0.5842, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.12062256809338522, |
| "grad_norm": 0.9504477977752686, |
| "learning_rate": 0.00010545454545454545, |
| "loss": 0.5995, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.1245136186770428, |
| "grad_norm": 1.7914175987243652, |
| "learning_rate": 0.00010181818181818181, |
| "loss": 0.5761, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.12840466926070038, |
| "grad_norm": 0.7228565812110901, |
| "learning_rate": 9.818181818181818e-05, |
| "loss": 0.5816, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.13229571984435798, |
| "grad_norm": 0.5315433740615845, |
| "learning_rate": 9.454545454545455e-05, |
| "loss": 0.5408, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.13618677042801555, |
| "grad_norm": 0.5710902214050293, |
| "learning_rate": 9.090909090909092e-05, |
| "loss": 0.5474, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.14007782101167315, |
| "grad_norm": 0.5707881450653076, |
| "learning_rate": 8.727272727272727e-05, |
| "loss": 0.5588, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.14396887159533073, |
| "grad_norm": 0.6319161653518677, |
| "learning_rate": 8.363636363636364e-05, |
| "loss": 0.5617, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.14785992217898833, |
| "grad_norm": 0.6077592372894287, |
| "learning_rate": 8e-05, |
| "loss": 0.5928, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.1517509727626459, |
| "grad_norm": 0.5045843720436096, |
| "learning_rate": 7.636363636363637e-05, |
| "loss": 0.4255, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.1556420233463035, |
| "grad_norm": 0.519895613193512, |
| "learning_rate": 7.272727272727273e-05, |
| "loss": 0.4984, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.15953307392996108, |
| "grad_norm": 0.7286504507064819, |
| "learning_rate": 6.90909090909091e-05, |
| "loss": 0.5935, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.16342412451361868, |
| "grad_norm": 0.5982980132102966, |
| "learning_rate": 6.545454545454546e-05, |
| "loss": 0.6141, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.16731517509727625, |
| "grad_norm": 0.5843263268470764, |
| "learning_rate": 6.181818181818182e-05, |
| "loss": 0.4612, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.17120622568093385, |
| "grad_norm": 0.6048635244369507, |
| "learning_rate": 5.818181818181818e-05, |
| "loss": 0.5251, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.17509727626459143, |
| "grad_norm": 0.7759015560150146, |
| "learning_rate": 5.4545454545454546e-05, |
| "loss": 0.5338, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.17898832684824903, |
| "grad_norm": 0.6101873517036438, |
| "learning_rate": 5.090909090909091e-05, |
| "loss": 0.6087, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.1828793774319066, |
| "grad_norm": 0.714209258556366, |
| "learning_rate": 4.7272727272727275e-05, |
| "loss": 0.581, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.1867704280155642, |
| "grad_norm": 0.6215356588363647, |
| "learning_rate": 4.3636363636363636e-05, |
| "loss": 0.6139, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.19066147859922178, |
| "grad_norm": 0.531689465045929, |
| "learning_rate": 4e-05, |
| "loss": 0.4751, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.19455252918287938, |
| "grad_norm": 0.705681562423706, |
| "learning_rate": 3.6363636363636364e-05, |
| "loss": 0.6481, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.19844357976653695, |
| "grad_norm": 0.613798201084137, |
| "learning_rate": 3.272727272727273e-05, |
| "loss": 0.6057, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.20233463035019456, |
| "grad_norm": 0.5866016149520874, |
| "learning_rate": 2.909090909090909e-05, |
| "loss": 0.5136, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.20622568093385213, |
| "grad_norm": 0.5885356664657593, |
| "learning_rate": 2.5454545454545454e-05, |
| "loss": 0.4433, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.21011673151750973, |
| "grad_norm": 0.5768156051635742, |
| "learning_rate": 2.1818181818181818e-05, |
| "loss": 0.5591, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.2140077821011673, |
| "grad_norm": 0.7024815678596497, |
| "learning_rate": 1.8181818181818182e-05, |
| "loss": 0.6333, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.2178988326848249, |
| "grad_norm": 0.5791760683059692, |
| "learning_rate": 1.4545454545454545e-05, |
| "loss": 0.4965, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.22178988326848248, |
| "grad_norm": 0.5099902153015137, |
| "learning_rate": 1.0909090909090909e-05, |
| "loss": 0.4959, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.22568093385214008, |
| "grad_norm": 0.5241119265556335, |
| "learning_rate": 7.272727272727272e-06, |
| "loss": 0.3799, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.22957198443579765, |
| "grad_norm": 0.6263752579689026, |
| "learning_rate": 3.636363636363636e-06, |
| "loss": 0.4662, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.23346303501945526, |
| "grad_norm": 0.6203947067260742, |
| "learning_rate": 0.0, |
| "loss": 0.5361, |
| "step": 60 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 60, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1679389378215936.0, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
