| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 78, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.038461538461538464, |
| "grad_norm": 6.870266914367676, |
| "learning_rate": 1.25e-06, |
| "loss": 1.1223, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.07692307692307693, |
| "grad_norm": 6.60228157043457, |
| "learning_rate": 2.5e-06, |
| "loss": 1.0548, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.11538461538461539, |
| "grad_norm": 6.515733242034912, |
| "learning_rate": 3.7500000000000005e-06, |
| "loss": 1.052, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.15384615384615385, |
| "grad_norm": 5.764657497406006, |
| "learning_rate": 5e-06, |
| "loss": 1.0166, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.19230769230769232, |
| "grad_norm": 4.684140205383301, |
| "learning_rate": 6.25e-06, |
| "loss": 0.9942, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.23076923076923078, |
| "grad_norm": 2.8065361976623535, |
| "learning_rate": 7.500000000000001e-06, |
| "loss": 0.9359, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.2692307692307692, |
| "grad_norm": 4.593384742736816, |
| "learning_rate": 8.750000000000001e-06, |
| "loss": 0.9657, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.3076923076923077, |
| "grad_norm": 4.550309181213379, |
| "learning_rate": 1e-05, |
| "loss": 0.9009, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.34615384615384615, |
| "grad_norm": 4.862745761871338, |
| "learning_rate": 9.994965332706574e-06, |
| "loss": 0.9548, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.38461538461538464, |
| "grad_norm": 4.1590962409973145, |
| "learning_rate": 9.979871469976197e-06, |
| "loss": 0.8451, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.4230769230769231, |
| "grad_norm": 3.6216330528259277, |
| "learning_rate": 9.954748808839675e-06, |
| "loss": 0.9311, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.46153846153846156, |
| "grad_norm": 2.3328566551208496, |
| "learning_rate": 9.91964794299315e-06, |
| "loss": 0.9269, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.5, |
| "grad_norm": 1.803200602531433, |
| "learning_rate": 9.874639560909118e-06, |
| "loss": 0.8337, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.5384615384615384, |
| "grad_norm": 1.7843149900436401, |
| "learning_rate": 9.819814303479268e-06, |
| "loss": 0.8367, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.5769230769230769, |
| "grad_norm": 1.253243327140808, |
| "learning_rate": 9.755282581475769e-06, |
| "loss": 0.769, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.6153846153846154, |
| "grad_norm": 1.2718814611434937, |
| "learning_rate": 9.681174353198687e-06, |
| "loss": 0.8519, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.6538461538461539, |
| "grad_norm": 1.3002601861953735, |
| "learning_rate": 9.597638862757255e-06, |
| "loss": 0.8114, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.6923076923076923, |
| "grad_norm": 1.0596836805343628, |
| "learning_rate": 9.504844339512096e-06, |
| "loss": 0.7983, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.7307692307692307, |
| "grad_norm": 1.2702919244766235, |
| "learning_rate": 9.40297765928369e-06, |
| "loss": 0.854, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.7692307692307693, |
| "grad_norm": 0.8957973122596741, |
| "learning_rate": 9.292243968009332e-06, |
| "loss": 0.7384, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.8076923076923077, |
| "grad_norm": 0.8740140199661255, |
| "learning_rate": 9.172866268606514e-06, |
| "loss": 0.8076, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.8461538461538461, |
| "grad_norm": 0.7290602922439575, |
| "learning_rate": 9.045084971874738e-06, |
| "loss": 0.7851, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.8846153846153846, |
| "grad_norm": 0.7917980551719666, |
| "learning_rate": 8.90915741234015e-06, |
| "loss": 0.8086, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.9230769230769231, |
| "grad_norm": 0.8550074696540833, |
| "learning_rate": 8.765357330018056e-06, |
| "loss": 0.7766, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.9615384615384616, |
| "grad_norm": 0.8657763004302979, |
| "learning_rate": 8.613974319136959e-06, |
| "loss": 0.8419, |
| "step": 25 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.7246716618537903, |
| "learning_rate": 8.455313244934324e-06, |
| "loss": 0.7583, |
| "step": 26 |
| }, |
| { |
| "epoch": 1.0384615384615385, |
| "grad_norm": 0.7091710567474365, |
| "learning_rate": 8.289693629698564e-06, |
| "loss": 0.7392, |
| "step": 27 |
| }, |
| { |
| "epoch": 1.0769230769230769, |
| "grad_norm": 0.7319567799568176, |
| "learning_rate": 8.117449009293668e-06, |
| "loss": 0.7661, |
| "step": 28 |
| }, |
| { |
| "epoch": 1.1153846153846154, |
| "grad_norm": 0.6999830007553101, |
| "learning_rate": 7.938926261462366e-06, |
| "loss": 0.7241, |
| "step": 29 |
| }, |
| { |
| "epoch": 1.1538461538461537, |
| "grad_norm": 0.637765109539032, |
| "learning_rate": 7.754484907260513e-06, |
| "loss": 0.7571, |
| "step": 30 |
| }, |
| { |
| "epoch": 1.1923076923076923, |
| "grad_norm": 0.7767643928527832, |
| "learning_rate": 7.564496387029532e-06, |
| "loss": 0.7495, |
| "step": 31 |
| }, |
| { |
| "epoch": 1.2307692307692308, |
| "grad_norm": 0.6210612654685974, |
| "learning_rate": 7.369343312364994e-06, |
| "loss": 0.7911, |
| "step": 32 |
| }, |
| { |
| "epoch": 1.2692307692307692, |
| "grad_norm": 0.600116491317749, |
| "learning_rate": 7.169418695587791e-06, |
| "loss": 0.7276, |
| "step": 33 |
| }, |
| { |
| "epoch": 1.3076923076923077, |
| "grad_norm": 0.6836248636245728, |
| "learning_rate": 6.965125158269619e-06, |
| "loss": 0.778, |
| "step": 34 |
| }, |
| { |
| "epoch": 1.3461538461538463, |
| "grad_norm": 0.6598909497261047, |
| "learning_rate": 6.7568741204067145e-06, |
| "loss": 0.761, |
| "step": 35 |
| }, |
| { |
| "epoch": 1.3846153846153846, |
| "grad_norm": 0.6266419887542725, |
| "learning_rate": 6.545084971874738e-06, |
| "loss": 0.7089, |
| "step": 36 |
| }, |
| { |
| "epoch": 1.4230769230769231, |
| "grad_norm": 0.6317741870880127, |
| "learning_rate": 6.330184227833376e-06, |
| "loss": 0.7458, |
| "step": 37 |
| }, |
| { |
| "epoch": 1.4615384615384617, |
| "grad_norm": 0.4944937825202942, |
| "learning_rate": 6.112604669781572e-06, |
| "loss": 0.7355, |
| "step": 38 |
| }, |
| { |
| "epoch": 1.5, |
| "grad_norm": 0.5044434666633606, |
| "learning_rate": 5.892784473993184e-06, |
| "loss": 0.7276, |
| "step": 39 |
| }, |
| { |
| "epoch": 1.5384615384615383, |
| "grad_norm": 0.6830148696899414, |
| "learning_rate": 5.671166329088278e-06, |
| "loss": 0.7389, |
| "step": 40 |
| }, |
| { |
| "epoch": 1.5769230769230769, |
| "grad_norm": 0.5602949857711792, |
| "learning_rate": 5.448196544517168e-06, |
| "loss": 0.7485, |
| "step": 41 |
| }, |
| { |
| "epoch": 1.6153846153846154, |
| "grad_norm": 0.516609787940979, |
| "learning_rate": 5.224324151752575e-06, |
| "loss": 0.7459, |
| "step": 42 |
| }, |
| { |
| "epoch": 1.6538461538461537, |
| "grad_norm": 0.5174584984779358, |
| "learning_rate": 5e-06, |
| "loss": 0.7106, |
| "step": 43 |
| }, |
| { |
| "epoch": 1.6923076923076923, |
| "grad_norm": 0.5429850816726685, |
| "learning_rate": 4.775675848247427e-06, |
| "loss": 0.7344, |
| "step": 44 |
| }, |
| { |
| "epoch": 1.7307692307692308, |
| "grad_norm": 0.5647832751274109, |
| "learning_rate": 4.551803455482833e-06, |
| "loss": 0.6985, |
| "step": 45 |
| }, |
| { |
| "epoch": 1.7692307692307692, |
| "grad_norm": 0.4936504662036896, |
| "learning_rate": 4.3288336709117246e-06, |
| "loss": 0.7706, |
| "step": 46 |
| }, |
| { |
| "epoch": 1.8076923076923077, |
| "grad_norm": 0.4863971471786499, |
| "learning_rate": 4.107215526006818e-06, |
| "loss": 0.7063, |
| "step": 47 |
| }, |
| { |
| "epoch": 1.8461538461538463, |
| "grad_norm": 0.4576539993286133, |
| "learning_rate": 3.887395330218429e-06, |
| "loss": 0.7289, |
| "step": 48 |
| }, |
| { |
| "epoch": 1.8846153846153846, |
| "grad_norm": 0.6159549355506897, |
| "learning_rate": 3.669815772166625e-06, |
| "loss": 0.6677, |
| "step": 49 |
| }, |
| { |
| "epoch": 1.9230769230769231, |
| "grad_norm": 0.48231181502342224, |
| "learning_rate": 3.4549150281252635e-06, |
| "loss": 0.7245, |
| "step": 50 |
| }, |
| { |
| "epoch": 1.9615384615384617, |
| "grad_norm": 0.4837266802787781, |
| "learning_rate": 3.2431258795932863e-06, |
| "loss": 0.6666, |
| "step": 51 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 0.5765633583068848, |
| "learning_rate": 3.0348748417303826e-06, |
| "loss": 0.7458, |
| "step": 52 |
| }, |
| { |
| "epoch": 2.0384615384615383, |
| "grad_norm": 0.44854843616485596, |
| "learning_rate": 2.83058130441221e-06, |
| "loss": 0.6414, |
| "step": 53 |
| }, |
| { |
| "epoch": 2.076923076923077, |
| "grad_norm": 0.4607446789741516, |
| "learning_rate": 2.6306566876350072e-06, |
| "loss": 0.665, |
| "step": 54 |
| }, |
| { |
| "epoch": 2.1153846153846154, |
| "grad_norm": 0.49104422330856323, |
| "learning_rate": 2.43550361297047e-06, |
| "loss": 0.708, |
| "step": 55 |
| }, |
| { |
| "epoch": 2.1538461538461537, |
| "grad_norm": 0.5487802624702454, |
| "learning_rate": 2.245515092739488e-06, |
| "loss": 0.6931, |
| "step": 56 |
| }, |
| { |
| "epoch": 2.1923076923076925, |
| "grad_norm": 0.5082237124443054, |
| "learning_rate": 2.061073738537635e-06, |
| "loss": 0.7527, |
| "step": 57 |
| }, |
| { |
| "epoch": 2.230769230769231, |
| "grad_norm": 0.3936366140842438, |
| "learning_rate": 1.8825509907063328e-06, |
| "loss": 0.6653, |
| "step": 58 |
| }, |
| { |
| "epoch": 2.269230769230769, |
| "grad_norm": 0.4543362259864807, |
| "learning_rate": 1.7103063703014372e-06, |
| "loss": 0.731, |
| "step": 59 |
| }, |
| { |
| "epoch": 2.3076923076923075, |
| "grad_norm": 0.4479061961174011, |
| "learning_rate": 1.544686755065677e-06, |
| "loss": 0.6771, |
| "step": 60 |
| }, |
| { |
| "epoch": 2.3461538461538463, |
| "grad_norm": 0.4659906029701233, |
| "learning_rate": 1.3860256808630429e-06, |
| "loss": 0.7394, |
| "step": 61 |
| }, |
| { |
| "epoch": 2.3846153846153846, |
| "grad_norm": 0.46474915742874146, |
| "learning_rate": 1.234642669981946e-06, |
| "loss": 0.6736, |
| "step": 62 |
| }, |
| { |
| "epoch": 2.423076923076923, |
| "grad_norm": 0.48529791831970215, |
| "learning_rate": 1.0908425876598512e-06, |
| "loss": 0.7392, |
| "step": 63 |
| }, |
| { |
| "epoch": 2.4615384615384617, |
| "grad_norm": 0.4053143560886383, |
| "learning_rate": 9.549150281252633e-07, |
| "loss": 0.7555, |
| "step": 64 |
| }, |
| { |
| "epoch": 2.5, |
| "grad_norm": 0.5247453451156616, |
| "learning_rate": 8.271337313934869e-07, |
| "loss": 0.7029, |
| "step": 65 |
| }, |
| { |
| "epoch": 2.5384615384615383, |
| "grad_norm": 0.3968586027622223, |
| "learning_rate": 7.077560319906696e-07, |
| "loss": 0.6822, |
| "step": 66 |
| }, |
| { |
| "epoch": 2.5769230769230766, |
| "grad_norm": 0.375961571931839, |
| "learning_rate": 5.9702234071631e-07, |
| "loss": 0.7254, |
| "step": 67 |
| }, |
| { |
| "epoch": 2.6153846153846154, |
| "grad_norm": 0.41553905606269836, |
| "learning_rate": 4.951556604879049e-07, |
| "loss": 0.7007, |
| "step": 68 |
| }, |
| { |
| "epoch": 2.6538461538461537, |
| "grad_norm": 0.38777419924736023, |
| "learning_rate": 4.0236113724274716e-07, |
| "loss": 0.6582, |
| "step": 69 |
| }, |
| { |
| "epoch": 2.6923076923076925, |
| "grad_norm": 0.41188523173332214, |
| "learning_rate": 3.18825646801314e-07, |
| "loss": 0.6747, |
| "step": 70 |
| }, |
| { |
| "epoch": 2.730769230769231, |
| "grad_norm": 0.4232975244522095, |
| "learning_rate": 2.447174185242324e-07, |
| "loss": 0.6905, |
| "step": 71 |
| }, |
| { |
| "epoch": 2.769230769230769, |
| "grad_norm": 0.4132767915725708, |
| "learning_rate": 1.801856965207338e-07, |
| "loss": 0.7099, |
| "step": 72 |
| }, |
| { |
| "epoch": 2.8076923076923075, |
| "grad_norm": 0.3853932023048401, |
| "learning_rate": 1.253604390908819e-07, |
| "loss": 0.6352, |
| "step": 73 |
| }, |
| { |
| "epoch": 2.8461538461538463, |
| "grad_norm": 0.4183899462223053, |
| "learning_rate": 8.035205700685167e-08, |
| "loss": 0.721, |
| "step": 74 |
| }, |
| { |
| "epoch": 2.8846153846153846, |
| "grad_norm": 0.4098947048187256, |
| "learning_rate": 4.52511911603265e-08, |
| "loss": 0.6852, |
| "step": 75 |
| }, |
| { |
| "epoch": 2.9230769230769234, |
| "grad_norm": 0.39528220891952515, |
| "learning_rate": 2.012853002380466e-08, |
| "loss": 0.6431, |
| "step": 76 |
| }, |
| { |
| "epoch": 2.9615384615384617, |
| "grad_norm": 0.406534343957901, |
| "learning_rate": 5.034667293427053e-09, |
| "loss": 0.7168, |
| "step": 77 |
| }, |
| { |
| "epoch": 3.0, |
| "grad_norm": 0.35006237030029297, |
| "learning_rate": 0.0, |
| "loss": 0.6841, |
| "step": 78 |
| }, |
| { |
| "epoch": 3.0, |
| "step": 78, |
| "total_flos": 73738752491520.0, |
| "train_loss": 0.7710517148176829, |
| "train_runtime": 7635.6931, |
| "train_samples_per_second": 0.981, |
| "train_steps_per_second": 0.01 |
| } |
| ], |
| "logging_steps": 1.0, |
| "max_steps": 78, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "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": 73738752491520.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
