| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 6.666666666666667, |
| "eval_steps": 500, |
| "global_step": 70, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.09523809523809523, |
| "grad_norm": 5.826747964736604, |
| "learning_rate": 2.8571428571428573e-06, |
| "loss": 0.9545, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.19047619047619047, |
| "grad_norm": 6.284509468268676, |
| "learning_rate": 5.7142857142857145e-06, |
| "loss": 0.9963, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.2857142857142857, |
| "grad_norm": 5.8099958656024295, |
| "learning_rate": 8.571428571428571e-06, |
| "loss": 0.9956, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.38095238095238093, |
| "grad_norm": 2.2399326619468747, |
| "learning_rate": 1.1428571428571429e-05, |
| "loss": 0.9425, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.47619047619047616, |
| "grad_norm": 4.651161883322413, |
| "learning_rate": 1.4285714285714287e-05, |
| "loss": 0.9215, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.5714285714285714, |
| "grad_norm": 4.875461531398872, |
| "learning_rate": 1.7142857142857142e-05, |
| "loss": 0.8968, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.6666666666666666, |
| "grad_norm": 4.852597430322658, |
| "learning_rate": 2e-05, |
| "loss": 0.8699, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.7619047619047619, |
| "grad_norm": 3.650941884259547, |
| "learning_rate": 1.9987569212189224e-05, |
| "loss": 0.8294, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.8571428571428571, |
| "grad_norm": 3.5195078895869676, |
| "learning_rate": 1.9950307753654016e-05, |
| "loss": 0.8498, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.9523809523809523, |
| "grad_norm": 2.572471476791739, |
| "learning_rate": 1.9888308262251286e-05, |
| "loss": 0.7961, |
| "step": 10 |
| }, |
| { |
| "epoch": 1.0476190476190477, |
| "grad_norm": 3.179087350836649, |
| "learning_rate": 1.9801724878485438e-05, |
| "loss": 1.143, |
| "step": 11 |
| }, |
| { |
| "epoch": 1.1428571428571428, |
| "grad_norm": 1.5221610360522788, |
| "learning_rate": 1.969077286229078e-05, |
| "loss": 0.7095, |
| "step": 12 |
| }, |
| { |
| "epoch": 1.2380952380952381, |
| "grad_norm": 1.7000333386361333, |
| "learning_rate": 1.955572805786141e-05, |
| "loss": 0.7578, |
| "step": 13 |
| }, |
| { |
| "epoch": 1.3333333333333333, |
| "grad_norm": 1.2039602232048938, |
| "learning_rate": 1.9396926207859085e-05, |
| "loss": 0.6678, |
| "step": 14 |
| }, |
| { |
| "epoch": 1.4285714285714286, |
| "grad_norm": 1.0174143782433702, |
| "learning_rate": 1.921476211870408e-05, |
| "loss": 0.7232, |
| "step": 15 |
| }, |
| { |
| "epoch": 1.5238095238095237, |
| "grad_norm": 1.179150194851723, |
| "learning_rate": 1.900968867902419e-05, |
| "loss": 0.7196, |
| "step": 16 |
| }, |
| { |
| "epoch": 1.619047619047619, |
| "grad_norm": 0.7720644071863747, |
| "learning_rate": 1.8782215733702286e-05, |
| "loss": 0.7044, |
| "step": 17 |
| }, |
| { |
| "epoch": 1.7142857142857144, |
| "grad_norm": 0.7515363216015968, |
| "learning_rate": 1.8532908816321557e-05, |
| "loss": 0.6598, |
| "step": 18 |
| }, |
| { |
| "epoch": 1.8095238095238095, |
| "grad_norm": 0.7116717493477911, |
| "learning_rate": 1.826238774315995e-05, |
| "loss": 0.6985, |
| "step": 19 |
| }, |
| { |
| "epoch": 1.9047619047619047, |
| "grad_norm": 0.7748030640555293, |
| "learning_rate": 1.7971325072229227e-05, |
| "loss": 0.6762, |
| "step": 20 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 1.1248724935431351, |
| "learning_rate": 1.766044443118978e-05, |
| "loss": 0.944, |
| "step": 21 |
| }, |
| { |
| "epoch": 2.0952380952380953, |
| "grad_norm": 0.6637421291764616, |
| "learning_rate": 1.7330518718298263e-05, |
| "loss": 0.5937, |
| "step": 22 |
| }, |
| { |
| "epoch": 2.1904761904761907, |
| "grad_norm": 0.6288305084891433, |
| "learning_rate": 1.698236818086073e-05, |
| "loss": 0.5627, |
| "step": 23 |
| }, |
| { |
| "epoch": 2.2857142857142856, |
| "grad_norm": 0.7061563238885146, |
| "learning_rate": 1.6616858375968596e-05, |
| "loss": 0.5685, |
| "step": 24 |
| }, |
| { |
| "epoch": 2.380952380952381, |
| "grad_norm": 0.6626965340895049, |
| "learning_rate": 1.6234898018587336e-05, |
| "loss": 0.5741, |
| "step": 25 |
| }, |
| { |
| "epoch": 2.4761904761904763, |
| "grad_norm": 0.7250182630915804, |
| "learning_rate": 1.5837436722347902e-05, |
| "loss": 0.5585, |
| "step": 26 |
| }, |
| { |
| "epoch": 2.571428571428571, |
| "grad_norm": 0.7587941022271201, |
| "learning_rate": 1.5425462638657597e-05, |
| "loss": 0.578, |
| "step": 27 |
| }, |
| { |
| "epoch": 2.6666666666666665, |
| "grad_norm": 0.6218747796395786, |
| "learning_rate": 1.5000000000000002e-05, |
| "loss": 0.5398, |
| "step": 28 |
| }, |
| { |
| "epoch": 2.761904761904762, |
| "grad_norm": 0.7330380047350604, |
| "learning_rate": 1.4562106573531632e-05, |
| "loss": 0.5607, |
| "step": 29 |
| }, |
| { |
| "epoch": 2.857142857142857, |
| "grad_norm": 0.7342275810894874, |
| "learning_rate": 1.4112871031306118e-05, |
| "loss": 0.5327, |
| "step": 30 |
| }, |
| { |
| "epoch": 2.9523809523809526, |
| "grad_norm": 0.7117726792634246, |
| "learning_rate": 1.3653410243663953e-05, |
| "loss": 0.5451, |
| "step": 31 |
| }, |
| { |
| "epoch": 3.0476190476190474, |
| "grad_norm": 0.9947018101140755, |
| "learning_rate": 1.3184866502516846e-05, |
| "loss": 0.7337, |
| "step": 32 |
| }, |
| { |
| "epoch": 3.142857142857143, |
| "grad_norm": 0.7630649127419351, |
| "learning_rate": 1.2708404681430054e-05, |
| "loss": 0.4906, |
| "step": 33 |
| }, |
| { |
| "epoch": 3.238095238095238, |
| "grad_norm": 0.5854869504094958, |
| "learning_rate": 1.2225209339563144e-05, |
| "loss": 0.4505, |
| "step": 34 |
| }, |
| { |
| "epoch": 3.3333333333333335, |
| "grad_norm": 0.7590129849650509, |
| "learning_rate": 1.1736481776669307e-05, |
| "loss": 0.4561, |
| "step": 35 |
| }, |
| { |
| "epoch": 3.4285714285714284, |
| "grad_norm": 0.5575466419744181, |
| "learning_rate": 1.1243437046474854e-05, |
| "loss": 0.4272, |
| "step": 36 |
| }, |
| { |
| "epoch": 3.5238095238095237, |
| "grad_norm": 0.6923046563052673, |
| "learning_rate": 1.0747300935864245e-05, |
| "loss": 0.4469, |
| "step": 37 |
| }, |
| { |
| "epoch": 3.619047619047619, |
| "grad_norm": 0.6797225969476902, |
| "learning_rate": 1.0249306917380731e-05, |
| "loss": 0.4163, |
| "step": 38 |
| }, |
| { |
| "epoch": 3.7142857142857144, |
| "grad_norm": 0.5385503945680159, |
| "learning_rate": 9.750693082619274e-06, |
| "loss": 0.4217, |
| "step": 39 |
| }, |
| { |
| "epoch": 3.8095238095238093, |
| "grad_norm": 0.6177047166400342, |
| "learning_rate": 9.252699064135759e-06, |
| "loss": 0.4681, |
| "step": 40 |
| }, |
| { |
| "epoch": 3.9047619047619047, |
| "grad_norm": 0.5341256640524201, |
| "learning_rate": 8.756562953525151e-06, |
| "loss": 0.4479, |
| "step": 41 |
| }, |
| { |
| "epoch": 4.0, |
| "grad_norm": 0.7935477073957191, |
| "learning_rate": 8.263518223330698e-06, |
| "loss": 0.6286, |
| "step": 42 |
| }, |
| { |
| "epoch": 4.095238095238095, |
| "grad_norm": 0.517842073614362, |
| "learning_rate": 7.774790660436857e-06, |
| "loss": 0.3714, |
| "step": 43 |
| }, |
| { |
| "epoch": 4.190476190476191, |
| "grad_norm": 0.538862773168787, |
| "learning_rate": 7.291595318569951e-06, |
| "loss": 0.3794, |
| "step": 44 |
| }, |
| { |
| "epoch": 4.285714285714286, |
| "grad_norm": 0.5022314928645712, |
| "learning_rate": 6.815133497483157e-06, |
| "loss": 0.3869, |
| "step": 45 |
| }, |
| { |
| "epoch": 4.380952380952381, |
| "grad_norm": 0.43924957754934535, |
| "learning_rate": 6.34658975633605e-06, |
| "loss": 0.3488, |
| "step": 46 |
| }, |
| { |
| "epoch": 4.476190476190476, |
| "grad_norm": 0.5225603374352963, |
| "learning_rate": 5.887128968693887e-06, |
| "loss": 0.3458, |
| "step": 47 |
| }, |
| { |
| "epoch": 4.571428571428571, |
| "grad_norm": 0.5240648864878037, |
| "learning_rate": 5.43789342646837e-06, |
| "loss": 0.3633, |
| "step": 48 |
| }, |
| { |
| "epoch": 4.666666666666667, |
| "grad_norm": 0.49098106557625276, |
| "learning_rate": 5.000000000000003e-06, |
| "loss": 0.368, |
| "step": 49 |
| }, |
| { |
| "epoch": 4.761904761904762, |
| "grad_norm": 0.4343098354303779, |
| "learning_rate": 4.5745373613424075e-06, |
| "loss": 0.3478, |
| "step": 50 |
| }, |
| { |
| "epoch": 4.857142857142857, |
| "grad_norm": 0.4618142447309536, |
| "learning_rate": 4.162563277652104e-06, |
| "loss": 0.3541, |
| "step": 51 |
| }, |
| { |
| "epoch": 4.9523809523809526, |
| "grad_norm": 0.43485648510875424, |
| "learning_rate": 3.7651019814126656e-06, |
| "loss": 0.3483, |
| "step": 52 |
| }, |
| { |
| "epoch": 5.0476190476190474, |
| "grad_norm": 0.7618906280322914, |
| "learning_rate": 3.3831416240314085e-06, |
| "loss": 0.5051, |
| "step": 53 |
| }, |
| { |
| "epoch": 5.142857142857143, |
| "grad_norm": 0.48563622277205065, |
| "learning_rate": 3.017631819139273e-06, |
| "loss": 0.3073, |
| "step": 54 |
| }, |
| { |
| "epoch": 5.238095238095238, |
| "grad_norm": 0.40906446587129447, |
| "learning_rate": 2.669481281701739e-06, |
| "loss": 0.3209, |
| "step": 55 |
| }, |
| { |
| "epoch": 5.333333333333333, |
| "grad_norm": 0.45758261233645986, |
| "learning_rate": 2.339555568810221e-06, |
| "loss": 0.3349, |
| "step": 56 |
| }, |
| { |
| "epoch": 5.428571428571429, |
| "grad_norm": 0.5038076463420655, |
| "learning_rate": 2.0286749277707783e-06, |
| "loss": 0.3231, |
| "step": 57 |
| }, |
| { |
| "epoch": 5.523809523809524, |
| "grad_norm": 0.5007063947430315, |
| "learning_rate": 1.7376122568400533e-06, |
| "loss": 0.3176, |
| "step": 58 |
| }, |
| { |
| "epoch": 5.619047619047619, |
| "grad_norm": 0.4743582895495735, |
| "learning_rate": 1.467091183678444e-06, |
| "loss": 0.318, |
| "step": 59 |
| }, |
| { |
| "epoch": 5.714285714285714, |
| "grad_norm": 0.39376361530852877, |
| "learning_rate": 1.2177842662977136e-06, |
| "loss": 0.289, |
| "step": 60 |
| }, |
| { |
| "epoch": 5.809523809523809, |
| "grad_norm": 0.39667308482069435, |
| "learning_rate": 9.903113209758098e-07, |
| "loss": 0.3285, |
| "step": 61 |
| }, |
| { |
| "epoch": 5.904761904761905, |
| "grad_norm": 0.3815656843384921, |
| "learning_rate": 7.852378812959227e-07, |
| "loss": 0.2843, |
| "step": 62 |
| }, |
| { |
| "epoch": 6.0, |
| "grad_norm": 0.6931498343995814, |
| "learning_rate": 6.030737921409169e-07, |
| "loss": 0.4462, |
| "step": 63 |
| }, |
| { |
| "epoch": 6.095238095238095, |
| "grad_norm": 0.4069309903695016, |
| "learning_rate": 4.4427194213859216e-07, |
| "loss": 0.309, |
| "step": 64 |
| }, |
| { |
| "epoch": 6.190476190476191, |
| "grad_norm": 0.40684312272506473, |
| "learning_rate": 3.0922713770922155e-07, |
| "loss": 0.302, |
| "step": 65 |
| }, |
| { |
| "epoch": 6.285714285714286, |
| "grad_norm": 0.3909098350044091, |
| "learning_rate": 1.9827512151456175e-07, |
| "loss": 0.2962, |
| "step": 66 |
| }, |
| { |
| "epoch": 6.380952380952381, |
| "grad_norm": 0.41074503211559577, |
| "learning_rate": 1.1169173774871478e-07, |
| "loss": 0.2891, |
| "step": 67 |
| }, |
| { |
| "epoch": 6.476190476190476, |
| "grad_norm": 0.4304153764224256, |
| "learning_rate": 4.9692246345985905e-08, |
| "loss": 0.3033, |
| "step": 68 |
| }, |
| { |
| "epoch": 6.571428571428571, |
| "grad_norm": 0.363241859642786, |
| "learning_rate": 1.2430787810776556e-08, |
| "loss": 0.2833, |
| "step": 69 |
| }, |
| { |
| "epoch": 6.666666666666667, |
| "grad_norm": 0.39006713929357784, |
| "learning_rate": 0.0, |
| "loss": 0.3071, |
| "step": 70 |
| }, |
| { |
| "epoch": 6.666666666666667, |
| "step": 70, |
| "total_flos": 1.0722729787392e+17, |
| "train_loss": 0.541948646732739, |
| "train_runtime": 2129.402, |
| "train_samples_per_second": 3.287, |
| "train_steps_per_second": 0.033 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 70, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 7, |
| "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": 1.0722729787392e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
