| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.202247191011236, | |
| "eval_steps": 27, | |
| "global_step": 214, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.016853932584269662, | |
| "grad_norm": 2.9885776042938232, | |
| "learning_rate": 1.4814814814814815e-06, | |
| "loss": 0.6012, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.033707865168539325, | |
| "grad_norm": 3.184929132461548, | |
| "learning_rate": 2.962962962962963e-06, | |
| "loss": 0.7573, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05056179775280899, | |
| "grad_norm": 3.256159782409668, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.9212, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.06741573033707865, | |
| "grad_norm": 2.833339214324951, | |
| "learning_rate": 5.925925925925926e-06, | |
| "loss": 0.6117, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.08426966292134831, | |
| "grad_norm": 3.08292818069458, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "loss": 0.8545, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "grad_norm": 2.317431688308716, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.6515, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.11797752808988764, | |
| "grad_norm": 2.9611644744873047, | |
| "learning_rate": 1.037037037037037e-05, | |
| "loss": 0.7159, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1348314606741573, | |
| "grad_norm": 2.698537826538086, | |
| "learning_rate": 1.1851851851851852e-05, | |
| "loss": 0.7019, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "grad_norm": 2.222154378890991, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.4411, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.578125, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.7817381620407104, | |
| "eval_VitaminC_cosine_ap": 0.5507972943944112, | |
| "eval_VitaminC_cosine_f1": 0.6595174262734584, | |
| "eval_VitaminC_cosine_f1_threshold": 0.28573715686798096, | |
| "eval_VitaminC_cosine_precision": 0.492, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.5703125, | |
| "eval_VitaminC_dot_accuracy_threshold": 316.7283020019531, | |
| "eval_VitaminC_dot_ap": 0.5511866185449577, | |
| "eval_VitaminC_dot_f1": 0.6577540106951871, | |
| "eval_VitaminC_dot_f1_threshold": 106.75863647460938, | |
| "eval_VitaminC_dot_precision": 0.4900398406374502, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.578125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 13.298419952392578, | |
| "eval_VitaminC_euclidean_ap": 0.5476323986807207, | |
| "eval_VitaminC_euclidean_f1": 0.6577540106951871, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.83933448791504, | |
| "eval_VitaminC_euclidean_precision": 0.4900398406374502, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.578125, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 279.69085693359375, | |
| "eval_VitaminC_manhattan_ap": 0.5412538781107805, | |
| "eval_VitaminC_manhattan_f1": 0.6577540106951871, | |
| "eval_VitaminC_manhattan_f1_threshold": 499.8836364746094, | |
| "eval_VitaminC_manhattan_precision": 0.4900398406374502, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.578125, | |
| "eval_VitaminC_max_accuracy_threshold": 316.7283020019531, | |
| "eval_VitaminC_max_ap": 0.5511866185449577, | |
| "eval_VitaminC_max_f1": 0.6595174262734584, | |
| "eval_VitaminC_max_f1_threshold": 499.8836364746094, | |
| "eval_VitaminC_max_precision": 0.492, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5511866185449577, | |
| "eval_sts-test_pearson_cosine": 0.8488243436029344, | |
| "eval_sts-test_pearson_dot": 0.8480167969551653, | |
| "eval_sts-test_pearson_euclidean": 0.8800283985117625, | |
| "eval_sts-test_pearson_manhattan": 0.880588311422627, | |
| "eval_sts-test_pearson_max": 0.880588311422627, | |
| "eval_sts-test_spearman_cosine": 0.8905659331642088, | |
| "eval_sts-test_spearman_dot": 0.8692084657204004, | |
| "eval_sts-test_spearman_euclidean": 0.8809566840232712, | |
| "eval_sts-test_spearman_manhattan": 0.883434007028195, | |
| "eval_sts-test_spearman_max": 0.8905659331642088, | |
| "eval_vitaminc-pairs_loss": 2.465860366821289, | |
| "eval_vitaminc-pairs_runtime": 1.4615, | |
| "eval_vitaminc-pairs_samples_per_second": 73.899, | |
| "eval_vitaminc-pairs_steps_per_second": 1.368, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_negation-triplets_loss": 1.7310789823532104, | |
| "eval_negation-triplets_runtime": 0.3009, | |
| "eval_negation-triplets_samples_per_second": 212.692, | |
| "eval_negation-triplets_steps_per_second": 3.323, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_scitail-pairs-pos_loss": 0.1150394082069397, | |
| "eval_scitail-pairs-pos_runtime": 0.3739, | |
| "eval_scitail-pairs-pos_samples_per_second": 144.431, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.675, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_xsum-pairs_loss": 0.11168850213289261, | |
| "eval_xsum-pairs_runtime": 3.1697, | |
| "eval_xsum-pairs_samples_per_second": 40.382, | |
| "eval_xsum-pairs_steps_per_second": 0.631, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_sciq_pairs_loss": 0.03450964391231537, | |
| "eval_sciq_pairs_runtime": 3.3283, | |
| "eval_sciq_pairs_samples_per_second": 38.459, | |
| "eval_sciq_pairs_steps_per_second": 0.601, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_qasc_pairs_loss": 0.11095743626356125, | |
| "eval_qasc_pairs_runtime": 0.6261, | |
| "eval_qasc_pairs_samples_per_second": 204.45, | |
| "eval_qasc_pairs_steps_per_second": 3.195, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_openbookqa_pairs_loss": 0.7092063426971436, | |
| "eval_openbookqa_pairs_runtime": 0.5866, | |
| "eval_openbookqa_pairs_samples_per_second": 218.19, | |
| "eval_openbookqa_pairs_steps_per_second": 3.409, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_msmarco_pairs_loss": 0.3955218493938446, | |
| "eval_msmarco_pairs_runtime": 1.2942, | |
| "eval_msmarco_pairs_samples_per_second": 98.902, | |
| "eval_msmarco_pairs_steps_per_second": 1.545, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_nq_pairs_loss": 0.42051073908805847, | |
| "eval_nq_pairs_runtime": 2.3875, | |
| "eval_nq_pairs_samples_per_second": 53.612, | |
| "eval_nq_pairs_steps_per_second": 0.838, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_trivia_pairs_loss": 0.93178790807724, | |
| "eval_trivia_pairs_runtime": 4.4363, | |
| "eval_trivia_pairs_samples_per_second": 28.853, | |
| "eval_trivia_pairs_steps_per_second": 0.451, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_gooaq_pairs_loss": 0.6505913138389587, | |
| "eval_gooaq_pairs_runtime": 0.8826, | |
| "eval_gooaq_pairs_samples_per_second": 145.027, | |
| "eval_gooaq_pairs_steps_per_second": 2.266, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15168539325842698, | |
| "eval_paws-pos_loss": 0.024931101128458977, | |
| "eval_paws-pos_runtime": 0.6852, | |
| "eval_paws-pos_samples_per_second": 186.805, | |
| "eval_paws-pos_steps_per_second": 2.919, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.16853932584269662, | |
| "grad_norm": 2.826900005340576, | |
| "learning_rate": 1.4814814814814815e-05, | |
| "loss": 0.5125, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1853932584269663, | |
| "grad_norm": 2.9938910007476807, | |
| "learning_rate": 1.6296296296296297e-05, | |
| "loss": 0.6885, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "grad_norm": 3.3046395778656006, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 0.6435, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.21910112359550563, | |
| "grad_norm": 2.4184651374816895, | |
| "learning_rate": 1.925925925925926e-05, | |
| "loss": 0.753, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.23595505617977527, | |
| "grad_norm": 2.9905433654785156, | |
| "learning_rate": 2.074074074074074e-05, | |
| "loss": 0.7427, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.25280898876404495, | |
| "grad_norm": 2.745820999145508, | |
| "learning_rate": 2.2222222222222227e-05, | |
| "loss": 0.5083, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.2696629213483146, | |
| "grad_norm": 2.6370577812194824, | |
| "learning_rate": 2.3703703703703703e-05, | |
| "loss": 0.7454, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.28651685393258425, | |
| "grad_norm": 3.044011116027832, | |
| "learning_rate": 2.5185185185185187e-05, | |
| "loss": 0.8356, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "grad_norm": 3.718804121017456, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.8864, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.57421875, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.7991844415664673, | |
| "eval_VitaminC_cosine_ap": 0.5485498837322925, | |
| "eval_VitaminC_cosine_f1": 0.6595174262734584, | |
| "eval_VitaminC_cosine_f1_threshold": 0.3160865008831024, | |
| "eval_VitaminC_cosine_precision": 0.492, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.578125, | |
| "eval_VitaminC_dot_accuracy_threshold": 327.0416564941406, | |
| "eval_VitaminC_dot_ap": 0.54993134882601, | |
| "eval_VitaminC_dot_f1": 0.6595174262734584, | |
| "eval_VitaminC_dot_f1_threshold": 117.44181060791016, | |
| "eval_VitaminC_dot_precision": 0.492, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.57421875, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 13.019258499145508, | |
| "eval_VitaminC_euclidean_ap": 0.5435066540334542, | |
| "eval_VitaminC_euclidean_f1": 0.6577540106951871, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.688644409179688, | |
| "eval_VitaminC_euclidean_precision": 0.4900398406374502, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.57421875, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 283.876220703125, | |
| "eval_VitaminC_manhattan_ap": 0.5416615397828658, | |
| "eval_VitaminC_manhattan_f1": 0.6559999999999999, | |
| "eval_VitaminC_manhattan_f1_threshold": 514.0216064453125, | |
| "eval_VitaminC_manhattan_precision": 0.4880952380952381, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.578125, | |
| "eval_VitaminC_max_accuracy_threshold": 327.0416564941406, | |
| "eval_VitaminC_max_ap": 0.54993134882601, | |
| "eval_VitaminC_max_f1": 0.6595174262734584, | |
| "eval_VitaminC_max_f1_threshold": 514.0216064453125, | |
| "eval_VitaminC_max_precision": 0.492, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.54993134882601, | |
| "eval_sts-test_pearson_cosine": 0.8452615878553369, | |
| "eval_sts-test_pearson_dot": 0.8404858620687519, | |
| "eval_sts-test_pearson_euclidean": 0.8780527810910925, | |
| "eval_sts-test_pearson_manhattan": 0.878916157345712, | |
| "eval_sts-test_pearson_max": 0.878916157345712, | |
| "eval_sts-test_spearman_cosine": 0.8876915367075635, | |
| "eval_sts-test_spearman_dot": 0.8608104875327304, | |
| "eval_sts-test_spearman_euclidean": 0.8804138856889071, | |
| "eval_sts-test_spearman_manhattan": 0.8822803815444743, | |
| "eval_sts-test_spearman_max": 0.8876915367075635, | |
| "eval_vitaminc-pairs_loss": 2.454524040222168, | |
| "eval_vitaminc-pairs_runtime": 1.4583, | |
| "eval_vitaminc-pairs_samples_per_second": 74.057, | |
| "eval_vitaminc-pairs_steps_per_second": 1.371, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_negation-triplets_loss": 1.7277792692184448, | |
| "eval_negation-triplets_runtime": 0.3027, | |
| "eval_negation-triplets_samples_per_second": 211.436, | |
| "eval_negation-triplets_steps_per_second": 3.304, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_scitail-pairs-pos_loss": 0.11168555170297623, | |
| "eval_scitail-pairs-pos_runtime": 0.3726, | |
| "eval_scitail-pairs-pos_samples_per_second": 144.911, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.684, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_xsum-pairs_loss": 0.10087604075670242, | |
| "eval_xsum-pairs_runtime": 3.1701, | |
| "eval_xsum-pairs_samples_per_second": 40.377, | |
| "eval_xsum-pairs_steps_per_second": 0.631, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_sciq_pairs_loss": 0.03466618433594704, | |
| "eval_sciq_pairs_runtime": 3.3778, | |
| "eval_sciq_pairs_samples_per_second": 37.895, | |
| "eval_sciq_pairs_steps_per_second": 0.592, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_qasc_pairs_loss": 0.10551701486110687, | |
| "eval_qasc_pairs_runtime": 0.6271, | |
| "eval_qasc_pairs_samples_per_second": 204.125, | |
| "eval_qasc_pairs_steps_per_second": 3.189, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_openbookqa_pairs_loss": 0.7239958643913269, | |
| "eval_openbookqa_pairs_runtime": 0.5811, | |
| "eval_openbookqa_pairs_samples_per_second": 220.255, | |
| "eval_openbookqa_pairs_steps_per_second": 3.441, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_msmarco_pairs_loss": 0.3808779716491699, | |
| "eval_msmarco_pairs_runtime": 1.2919, | |
| "eval_msmarco_pairs_samples_per_second": 99.082, | |
| "eval_msmarco_pairs_steps_per_second": 1.548, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_nq_pairs_loss": 0.44170400500297546, | |
| "eval_nq_pairs_runtime": 2.3835, | |
| "eval_nq_pairs_samples_per_second": 53.703, | |
| "eval_nq_pairs_steps_per_second": 0.839, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_trivia_pairs_loss": 0.9158428907394409, | |
| "eval_trivia_pairs_runtime": 4.4326, | |
| "eval_trivia_pairs_samples_per_second": 28.877, | |
| "eval_trivia_pairs_steps_per_second": 0.451, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_gooaq_pairs_loss": 0.6208247542381287, | |
| "eval_gooaq_pairs_runtime": 0.8797, | |
| "eval_gooaq_pairs_samples_per_second": 145.497, | |
| "eval_gooaq_pairs_steps_per_second": 2.273, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_paws-pos_loss": 0.02517784759402275, | |
| "eval_paws-pos_runtime": 0.694, | |
| "eval_paws-pos_samples_per_second": 184.442, | |
| "eval_paws-pos_steps_per_second": 2.882, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.3202247191011236, | |
| "grad_norm": 2.173736572265625, | |
| "learning_rate": 2.814814814814815e-05, | |
| "loss": 0.6015, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.33707865168539325, | |
| "grad_norm": 3.8964712619781494, | |
| "learning_rate": 2.962962962962963e-05, | |
| "loss": 0.9482, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3539325842696629, | |
| "grad_norm": 2.659498691558838, | |
| "learning_rate": 3.111111111111112e-05, | |
| "loss": 0.5404, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.3707865168539326, | |
| "grad_norm": 3.3499844074249268, | |
| "learning_rate": 3.259259259259259e-05, | |
| "loss": 0.805, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.38764044943820225, | |
| "grad_norm": 3.770142078399658, | |
| "learning_rate": 3.4074074074074077e-05, | |
| "loss": 0.7184, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "grad_norm": 3.740880012512207, | |
| "learning_rate": 3.555555555555555e-05, | |
| "loss": 0.8708, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.42134831460674155, | |
| "grad_norm": 2.981106996536255, | |
| "learning_rate": 3.703703703703704e-05, | |
| "loss": 0.8327, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.43820224719101125, | |
| "grad_norm": 2.3469011783599854, | |
| "learning_rate": 3.851851851851852e-05, | |
| "loss": 0.5025, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "grad_norm": 3.296035051345825, | |
| "learning_rate": 4e-05, | |
| "loss": 0.6517, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.578125, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.7859437465667725, | |
| "eval_VitaminC_cosine_ap": 0.5557444337961499, | |
| "eval_VitaminC_cosine_f1": 0.6595174262734584, | |
| "eval_VitaminC_cosine_f1_threshold": 0.3211573362350464, | |
| "eval_VitaminC_cosine_precision": 0.492, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.578125, | |
| "eval_VitaminC_dot_accuracy_threshold": 315.9444580078125, | |
| "eval_VitaminC_dot_ap": 0.5539524528858992, | |
| "eval_VitaminC_dot_f1": 0.6595174262734584, | |
| "eval_VitaminC_dot_f1_threshold": 129.88558959960938, | |
| "eval_VitaminC_dot_precision": 0.492, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.58203125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 13.113249778747559, | |
| "eval_VitaminC_euclidean_ap": 0.5510190217865811, | |
| "eval_VitaminC_euclidean_f1": 0.6577540106951871, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.90462303161621, | |
| "eval_VitaminC_euclidean_precision": 0.4900398406374502, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.578125, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 276.40142822265625, | |
| "eval_VitaminC_manhattan_ap": 0.5429240708188645, | |
| "eval_VitaminC_manhattan_f1": 0.6576819407008085, | |
| "eval_VitaminC_manhattan_f1_threshold": 469.7353515625, | |
| "eval_VitaminC_manhattan_precision": 0.49193548387096775, | |
| "eval_VitaminC_manhattan_recall": 0.991869918699187, | |
| "eval_VitaminC_max_accuracy": 0.58203125, | |
| "eval_VitaminC_max_accuracy_threshold": 315.9444580078125, | |
| "eval_VitaminC_max_ap": 0.5557444337961499, | |
| "eval_VitaminC_max_f1": 0.6595174262734584, | |
| "eval_VitaminC_max_f1_threshold": 469.7353515625, | |
| "eval_VitaminC_max_precision": 0.492, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5557444337961499, | |
| "eval_sts-test_pearson_cosine": 0.8483316632682467, | |
| "eval_sts-test_pearson_dot": 0.8392403098680445, | |
| "eval_sts-test_pearson_euclidean": 0.8814283057813619, | |
| "eval_sts-test_pearson_manhattan": 0.8815226866327923, | |
| "eval_sts-test_pearson_max": 0.8815226866327923, | |
| "eval_sts-test_spearman_cosine": 0.8903503892346, | |
| "eval_sts-test_spearman_dot": 0.857844431199042, | |
| "eval_sts-test_spearman_euclidean": 0.8851830636663006, | |
| "eval_sts-test_spearman_manhattan": 0.8865568876827619, | |
| "eval_sts-test_spearman_max": 0.8903503892346, | |
| "eval_vitaminc-pairs_loss": 2.3538782596588135, | |
| "eval_vitaminc-pairs_runtime": 1.4618, | |
| "eval_vitaminc-pairs_samples_per_second": 73.88, | |
| "eval_vitaminc-pairs_steps_per_second": 1.368, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_negation-triplets_loss": 1.649215579032898, | |
| "eval_negation-triplets_runtime": 0.3081, | |
| "eval_negation-triplets_samples_per_second": 207.723, | |
| "eval_negation-triplets_steps_per_second": 3.246, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_scitail-pairs-pos_loss": 0.11823470890522003, | |
| "eval_scitail-pairs-pos_runtime": 0.376, | |
| "eval_scitail-pairs-pos_samples_per_second": 143.616, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.66, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_xsum-pairs_loss": 0.08420603722333908, | |
| "eval_xsum-pairs_runtime": 3.1576, | |
| "eval_xsum-pairs_samples_per_second": 40.538, | |
| "eval_xsum-pairs_steps_per_second": 0.633, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_sciq_pairs_loss": 0.034781794995069504, | |
| "eval_sciq_pairs_runtime": 3.2597, | |
| "eval_sciq_pairs_samples_per_second": 39.267, | |
| "eval_sciq_pairs_steps_per_second": 0.614, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_qasc_pairs_loss": 0.10597346723079681, | |
| "eval_qasc_pairs_runtime": 0.6245, | |
| "eval_qasc_pairs_samples_per_second": 204.979, | |
| "eval_qasc_pairs_steps_per_second": 3.203, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_openbookqa_pairs_loss": 0.7160983681678772, | |
| "eval_openbookqa_pairs_runtime": 0.5767, | |
| "eval_openbookqa_pairs_samples_per_second": 221.961, | |
| "eval_openbookqa_pairs_steps_per_second": 3.468, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_msmarco_pairs_loss": 0.3454173803329468, | |
| "eval_msmarco_pairs_runtime": 1.2912, | |
| "eval_msmarco_pairs_samples_per_second": 99.134, | |
| "eval_msmarco_pairs_steps_per_second": 1.549, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_nq_pairs_loss": 0.4442503750324249, | |
| "eval_nq_pairs_runtime": 2.3854, | |
| "eval_nq_pairs_samples_per_second": 53.659, | |
| "eval_nq_pairs_steps_per_second": 0.838, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_trivia_pairs_loss": 0.9324482679367065, | |
| "eval_trivia_pairs_runtime": 4.4251, | |
| "eval_trivia_pairs_samples_per_second": 28.926, | |
| "eval_trivia_pairs_steps_per_second": 0.452, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_gooaq_pairs_loss": 0.6094165444374084, | |
| "eval_gooaq_pairs_runtime": 0.8751, | |
| "eval_gooaq_pairs_samples_per_second": 146.261, | |
| "eval_gooaq_pairs_steps_per_second": 2.285, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4550561797752809, | |
| "eval_paws-pos_loss": 0.024421451613307, | |
| "eval_paws-pos_runtime": 0.6865, | |
| "eval_paws-pos_samples_per_second": 186.444, | |
| "eval_paws-pos_steps_per_second": 2.913, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.47191011235955055, | |
| "grad_norm": 3.1395561695098877, | |
| "learning_rate": 3.999675367909485e-05, | |
| "loss": 0.5801, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.4887640449438202, | |
| "grad_norm": 2.7977917194366455, | |
| "learning_rate": 3.998701612152597e-05, | |
| "loss": 0.791, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "grad_norm": 2.3682048320770264, | |
| "learning_rate": 3.997079154212493e-05, | |
| "loss": 0.6042, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5224719101123596, | |
| "grad_norm": 2.843482255935669, | |
| "learning_rate": 3.99480869635839e-05, | |
| "loss": 0.7559, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5393258426966292, | |
| "grad_norm": 2.7346785068511963, | |
| "learning_rate": 3.9918912213415936e-05, | |
| "loss": 0.6258, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.5561797752808989, | |
| "grad_norm": 3.149007558822632, | |
| "learning_rate": 3.9883279919701226e-05, | |
| "loss": 0.8853, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.5730337078651685, | |
| "grad_norm": 3.3424761295318604, | |
| "learning_rate": 3.9841205505621106e-05, | |
| "loss": 0.5947, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.5898876404494382, | |
| "grad_norm": 2.6377146244049072, | |
| "learning_rate": 3.979270718278224e-05, | |
| "loss": 0.644, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "grad_norm": 1.3963145017623901, | |
| "learning_rate": 3.973780594333386e-05, | |
| "loss": 0.5682, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.58984375, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.7784540057182312, | |
| "eval_VitaminC_cosine_ap": 0.556890553952148, | |
| "eval_VitaminC_cosine_f1": 0.6595174262734584, | |
| "eval_VitaminC_cosine_f1_threshold": 0.3015836775302887, | |
| "eval_VitaminC_cosine_precision": 0.492, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.57421875, | |
| "eval_VitaminC_dot_accuracy_threshold": 310.07818603515625, | |
| "eval_VitaminC_dot_ap": 0.5486679382699982, | |
| "eval_VitaminC_dot_f1": 0.6595174262734584, | |
| "eval_VitaminC_dot_f1_threshold": 110.96945190429688, | |
| "eval_VitaminC_dot_precision": 0.492, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.578125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 13.410951614379883, | |
| "eval_VitaminC_euclidean_ap": 0.5551150763619972, | |
| "eval_VitaminC_euclidean_f1": 0.6577540106951871, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.933565139770508, | |
| "eval_VitaminC_euclidean_precision": 0.4900398406374502, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.58203125, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 236.87246704101562, | |
| "eval_VitaminC_manhattan_ap": 0.5465417933692003, | |
| "eval_VitaminC_manhattan_f1": 0.6576086956521738, | |
| "eval_VitaminC_manhattan_f1_threshold": 479.8819580078125, | |
| "eval_VitaminC_manhattan_precision": 0.49387755102040815, | |
| "eval_VitaminC_manhattan_recall": 0.983739837398374, | |
| "eval_VitaminC_max_accuracy": 0.58984375, | |
| "eval_VitaminC_max_accuracy_threshold": 310.07818603515625, | |
| "eval_VitaminC_max_ap": 0.556890553952148, | |
| "eval_VitaminC_max_f1": 0.6595174262734584, | |
| "eval_VitaminC_max_f1_threshold": 479.8819580078125, | |
| "eval_VitaminC_max_precision": 0.49387755102040815, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.556890553952148, | |
| "eval_sts-test_pearson_cosine": 0.8554011700695058, | |
| "eval_sts-test_pearson_dot": 0.8545651085608208, | |
| "eval_sts-test_pearson_euclidean": 0.8842988585732054, | |
| "eval_sts-test_pearson_manhattan": 0.8850809337540164, | |
| "eval_sts-test_pearson_max": 0.8850809337540164, | |
| "eval_sts-test_spearman_cosine": 0.8959079853731212, | |
| "eval_sts-test_spearman_dot": 0.876135947365041, | |
| "eval_sts-test_spearman_euclidean": 0.8856381406339634, | |
| "eval_sts-test_spearman_manhattan": 0.8868948834793577, | |
| "eval_sts-test_spearman_max": 0.8959079853731212, | |
| "eval_vitaminc-pairs_loss": 2.4271271228790283, | |
| "eval_vitaminc-pairs_runtime": 1.4672, | |
| "eval_vitaminc-pairs_samples_per_second": 73.61, | |
| "eval_vitaminc-pairs_steps_per_second": 1.363, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_negation-triplets_loss": 1.6174229383468628, | |
| "eval_negation-triplets_runtime": 0.3094, | |
| "eval_negation-triplets_samples_per_second": 206.869, | |
| "eval_negation-triplets_steps_per_second": 3.232, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_scitail-pairs-pos_loss": 0.10586681962013245, | |
| "eval_scitail-pairs-pos_runtime": 0.4307, | |
| "eval_scitail-pairs-pos_samples_per_second": 125.374, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.322, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_xsum-pairs_loss": 0.1041470319032669, | |
| "eval_xsum-pairs_runtime": 3.2912, | |
| "eval_xsum-pairs_samples_per_second": 38.891, | |
| "eval_xsum-pairs_steps_per_second": 0.608, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_sciq_pairs_loss": 0.03364330902695656, | |
| "eval_sciq_pairs_runtime": 3.3617, | |
| "eval_sciq_pairs_samples_per_second": 38.076, | |
| "eval_sciq_pairs_steps_per_second": 0.595, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_qasc_pairs_loss": 0.10827122628688812, | |
| "eval_qasc_pairs_runtime": 0.6238, | |
| "eval_qasc_pairs_samples_per_second": 205.206, | |
| "eval_qasc_pairs_steps_per_second": 3.206, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_openbookqa_pairs_loss": 0.7513518929481506, | |
| "eval_openbookqa_pairs_runtime": 0.5882, | |
| "eval_openbookqa_pairs_samples_per_second": 217.628, | |
| "eval_openbookqa_pairs_steps_per_second": 3.4, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_msmarco_pairs_loss": 0.3447520434856415, | |
| "eval_msmarco_pairs_runtime": 1.2968, | |
| "eval_msmarco_pairs_samples_per_second": 98.705, | |
| "eval_msmarco_pairs_steps_per_second": 1.542, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_nq_pairs_loss": 0.4307234287261963, | |
| "eval_nq_pairs_runtime": 2.4164, | |
| "eval_nq_pairs_samples_per_second": 52.971, | |
| "eval_nq_pairs_steps_per_second": 0.828, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_trivia_pairs_loss": 0.9260168075561523, | |
| "eval_trivia_pairs_runtime": 4.4644, | |
| "eval_trivia_pairs_samples_per_second": 28.671, | |
| "eval_trivia_pairs_steps_per_second": 0.448, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_gooaq_pairs_loss": 0.6301646828651428, | |
| "eval_gooaq_pairs_runtime": 0.9092, | |
| "eval_gooaq_pairs_samples_per_second": 140.777, | |
| "eval_gooaq_pairs_steps_per_second": 2.2, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_paws-pos_loss": 0.024637963622808456, | |
| "eval_paws-pos_runtime": 0.696, | |
| "eval_paws-pos_samples_per_second": 183.9, | |
| "eval_paws-pos_steps_per_second": 2.873, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6235955056179775, | |
| "grad_norm": 3.0150091648101807, | |
| "learning_rate": 3.9676525550881484e-05, | |
| "loss": 0.5974, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.6404494382022472, | |
| "grad_norm": 2.7985854148864746, | |
| "learning_rate": 3.9608892530200996e-05, | |
| "loss": 0.649, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.6573033707865169, | |
| "grad_norm": 3.00435471534729, | |
| "learning_rate": 3.953493615575757e-05, | |
| "loss": 0.6966, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.6741573033707865, | |
| "grad_norm": 3.276264190673828, | |
| "learning_rate": 3.945468843903448e-05, | |
| "loss": 0.542, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6910112359550562, | |
| "grad_norm": 3.144037961959839, | |
| "learning_rate": 3.936818411467709e-05, | |
| "loss": 0.8583, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "grad_norm": 2.498800754547119, | |
| "learning_rate": 3.9275460625458294e-05, | |
| "loss": 0.6416, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.7247191011235955, | |
| "grad_norm": 3.01488995552063, | |
| "learning_rate": 3.917655810607162e-05, | |
| "loss": 0.6273, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.7415730337078652, | |
| "grad_norm": 3.439838171005249, | |
| "learning_rate": 3.907151936575922e-05, | |
| "loss": 0.8621, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "grad_norm": 3.0246922969818115, | |
| "learning_rate": 3.896038986978224e-05, | |
| "loss": 0.7221, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.57421875, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8462294936180115, | |
| "eval_VitaminC_cosine_ap": 0.5610960529859609, | |
| "eval_VitaminC_cosine_f1": 0.6595174262734584, | |
| "eval_VitaminC_cosine_f1_threshold": 0.3176865577697754, | |
| "eval_VitaminC_cosine_precision": 0.492, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.5859375, | |
| "eval_VitaminC_dot_accuracy_threshold": 319.20416259765625, | |
| "eval_VitaminC_dot_ap": 0.5540963620104548, | |
| "eval_VitaminC_dot_f1": 0.6595174262734584, | |
| "eval_VitaminC_dot_f1_threshold": 121.18156433105469, | |
| "eval_VitaminC_dot_precision": 0.492, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.58203125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 15.303094863891602, | |
| "eval_VitaminC_euclidean_ap": 0.5590109366458975, | |
| "eval_VitaminC_euclidean_f1": 0.6577540106951871, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.605255126953125, | |
| "eval_VitaminC_euclidean_precision": 0.4900398406374502, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.578125, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 240.15660095214844, | |
| "eval_VitaminC_manhattan_ap": 0.5523438512618317, | |
| "eval_VitaminC_manhattan_f1": 0.6577540106951871, | |
| "eval_VitaminC_manhattan_f1_threshold": 503.4666748046875, | |
| "eval_VitaminC_manhattan_precision": 0.4900398406374502, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.5859375, | |
| "eval_VitaminC_max_accuracy_threshold": 319.20416259765625, | |
| "eval_VitaminC_max_ap": 0.5610960529859609, | |
| "eval_VitaminC_max_f1": 0.6595174262734584, | |
| "eval_VitaminC_max_f1_threshold": 503.4666748046875, | |
| "eval_VitaminC_max_precision": 0.492, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5610960529859609, | |
| "eval_sts-test_pearson_cosine": 0.8558158452024208, | |
| "eval_sts-test_pearson_dot": 0.8576514422982167, | |
| "eval_sts-test_pearson_euclidean": 0.8865043101846597, | |
| "eval_sts-test_pearson_manhattan": 0.8866527906896583, | |
| "eval_sts-test_pearson_max": 0.8866527906896583, | |
| "eval_sts-test_spearman_cosine": 0.8964919590428757, | |
| "eval_sts-test_spearman_dot": 0.8785377225806833, | |
| "eval_sts-test_spearman_euclidean": 0.8874999818863996, | |
| "eval_sts-test_spearman_manhattan": 0.8881272791699125, | |
| "eval_sts-test_spearman_max": 0.8964919590428757, | |
| "eval_vitaminc-pairs_loss": 2.336690902709961, | |
| "eval_vitaminc-pairs_runtime": 1.4561, | |
| "eval_vitaminc-pairs_samples_per_second": 74.172, | |
| "eval_vitaminc-pairs_steps_per_second": 1.374, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_negation-triplets_loss": 1.5712968111038208, | |
| "eval_negation-triplets_runtime": 0.2973, | |
| "eval_negation-triplets_samples_per_second": 215.304, | |
| "eval_negation-triplets_steps_per_second": 3.364, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_scitail-pairs-pos_loss": 0.1093834936618805, | |
| "eval_scitail-pairs-pos_runtime": 0.3709, | |
| "eval_scitail-pairs-pos_samples_per_second": 145.598, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.696, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_xsum-pairs_loss": 0.09297582507133484, | |
| "eval_xsum-pairs_runtime": 3.1496, | |
| "eval_xsum-pairs_samples_per_second": 40.64, | |
| "eval_xsum-pairs_steps_per_second": 0.635, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_sciq_pairs_loss": 0.031587302684783936, | |
| "eval_sciq_pairs_runtime": 3.2666, | |
| "eval_sciq_pairs_samples_per_second": 39.185, | |
| "eval_sciq_pairs_steps_per_second": 0.612, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_qasc_pairs_loss": 0.10210572183132172, | |
| "eval_qasc_pairs_runtime": 0.6258, | |
| "eval_qasc_pairs_samples_per_second": 204.538, | |
| "eval_qasc_pairs_steps_per_second": 3.196, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_openbookqa_pairs_loss": 0.7485109567642212, | |
| "eval_openbookqa_pairs_runtime": 0.5836, | |
| "eval_openbookqa_pairs_samples_per_second": 219.317, | |
| "eval_openbookqa_pairs_steps_per_second": 3.427, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_msmarco_pairs_loss": 0.34248754382133484, | |
| "eval_msmarco_pairs_runtime": 1.286, | |
| "eval_msmarco_pairs_samples_per_second": 99.53, | |
| "eval_msmarco_pairs_steps_per_second": 1.555, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_nq_pairs_loss": 0.41257673501968384, | |
| "eval_nq_pairs_runtime": 2.3755, | |
| "eval_nq_pairs_samples_per_second": 53.883, | |
| "eval_nq_pairs_steps_per_second": 0.842, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_trivia_pairs_loss": 0.9274640083312988, | |
| "eval_trivia_pairs_runtime": 4.4185, | |
| "eval_trivia_pairs_samples_per_second": 28.969, | |
| "eval_trivia_pairs_steps_per_second": 0.453, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_gooaq_pairs_loss": 0.5840359330177307, | |
| "eval_gooaq_pairs_runtime": 0.8784, | |
| "eval_gooaq_pairs_samples_per_second": 145.726, | |
| "eval_gooaq_pairs_steps_per_second": 2.277, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7584269662921348, | |
| "eval_paws-pos_loss": 0.024575484916567802, | |
| "eval_paws-pos_runtime": 0.6869, | |
| "eval_paws-pos_samples_per_second": 186.355, | |
| "eval_paws-pos_steps_per_second": 2.912, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.7752808988764045, | |
| "grad_norm": 3.189115285873413, | |
| "learning_rate": 3.884321771974146e-05, | |
| "loss": 0.9421, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.7921348314606742, | |
| "grad_norm": 1.8421012163162231, | |
| "learning_rate": 3.872005363275693e-05, | |
| "loss": 0.6845, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "grad_norm": 2.2710273265838623, | |
| "learning_rate": 3.859095091951534e-05, | |
| "loss": 0.5464, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.8258426966292135, | |
| "grad_norm": 2.98201060295105, | |
| "learning_rate": 3.845596546119496e-05, | |
| "loss": 0.6338, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.8426966292134831, | |
| "grad_norm": 2.9222068786621094, | |
| "learning_rate": 3.831515568527782e-05, | |
| "loss": 0.4993, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8595505617977528, | |
| "grad_norm": 3.6596696376800537, | |
| "learning_rate": 3.81685825402598e-05, | |
| "loss": 0.6939, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.8764044943820225, | |
| "grad_norm": 3.5363073348999023, | |
| "learning_rate": 3.801630946926956e-05, | |
| "loss": 0.5791, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.8932584269662921, | |
| "grad_norm": 3.6599326133728027, | |
| "learning_rate": 3.785840238260758e-05, | |
| "loss": 0.9226, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "grad_norm": 2.6967382431030273, | |
| "learning_rate": 3.7694929629217385e-05, | |
| "loss": 0.6336, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.578125, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8077100515365601, | |
| "eval_VitaminC_cosine_ap": 0.560345569715395, | |
| "eval_VitaminC_cosine_f1": 0.6577540106951871, | |
| "eval_VitaminC_cosine_f1_threshold": 0.25425243377685547, | |
| "eval_VitaminC_cosine_precision": 0.4900398406374502, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.578125, | |
| "eval_VitaminC_dot_accuracy_threshold": 331.0595703125, | |
| "eval_VitaminC_dot_ap": 0.5499174718252662, | |
| "eval_VitaminC_dot_f1": 0.6594594594594595, | |
| "eval_VitaminC_dot_f1_threshold": 121.5512924194336, | |
| "eval_VitaminC_dot_precision": 0.4939271255060729, | |
| "eval_VitaminC_dot_recall": 0.991869918699187, | |
| "eval_VitaminC_euclidean_accuracy": 0.58203125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 13.087348937988281, | |
| "eval_VitaminC_euclidean_ap": 0.5563988051869968, | |
| "eval_VitaminC_euclidean_f1": 0.6595174262734584, | |
| "eval_VitaminC_euclidean_f1_threshold": 24.085674285888672, | |
| "eval_VitaminC_euclidean_precision": 0.492, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.58203125, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 339.80157470703125, | |
| "eval_VitaminC_manhattan_ap": 0.5563552623581395, | |
| "eval_VitaminC_manhattan_f1": 0.6595174262734584, | |
| "eval_VitaminC_manhattan_f1_threshold": 513.6148681640625, | |
| "eval_VitaminC_manhattan_precision": 0.492, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.58203125, | |
| "eval_VitaminC_max_accuracy_threshold": 339.80157470703125, | |
| "eval_VitaminC_max_ap": 0.560345569715395, | |
| "eval_VitaminC_max_f1": 0.6595174262734584, | |
| "eval_VitaminC_max_f1_threshold": 513.6148681640625, | |
| "eval_VitaminC_max_precision": 0.4939271255060729, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.560345569715395, | |
| "eval_sts-test_pearson_cosine": 0.8572374144718338, | |
| "eval_sts-test_pearson_dot": 0.8574922304214377, | |
| "eval_sts-test_pearson_euclidean": 0.8855365860444931, | |
| "eval_sts-test_pearson_manhattan": 0.886788554614621, | |
| "eval_sts-test_pearson_max": 0.886788554614621, | |
| "eval_sts-test_spearman_cosine": 0.8965259734503722, | |
| "eval_sts-test_spearman_dot": 0.8784832995286892, | |
| "eval_sts-test_spearman_euclidean": 0.8855371715506584, | |
| "eval_sts-test_spearman_manhattan": 0.8887295132057965, | |
| "eval_sts-test_spearman_max": 0.8965259734503722, | |
| "eval_vitaminc-pairs_loss": 2.376066207885742, | |
| "eval_vitaminc-pairs_runtime": 1.4378, | |
| "eval_vitaminc-pairs_samples_per_second": 75.117, | |
| "eval_vitaminc-pairs_steps_per_second": 1.391, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_negation-triplets_loss": 1.63050377368927, | |
| "eval_negation-triplets_runtime": 0.3004, | |
| "eval_negation-triplets_samples_per_second": 213.075, | |
| "eval_negation-triplets_steps_per_second": 3.329, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_scitail-pairs-pos_loss": 0.10006655752658844, | |
| "eval_scitail-pairs-pos_runtime": 0.365, | |
| "eval_scitail-pairs-pos_samples_per_second": 147.952, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.74, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_xsum-pairs_loss": 0.0762382224202156, | |
| "eval_xsum-pairs_runtime": 3.1586, | |
| "eval_xsum-pairs_samples_per_second": 40.525, | |
| "eval_xsum-pairs_steps_per_second": 0.633, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_sciq_pairs_loss": 0.032084282487630844, | |
| "eval_sciq_pairs_runtime": 3.2586, | |
| "eval_sciq_pairs_samples_per_second": 39.28, | |
| "eval_sciq_pairs_steps_per_second": 0.614, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_qasc_pairs_loss": 0.10310036689043045, | |
| "eval_qasc_pairs_runtime": 0.6197, | |
| "eval_qasc_pairs_samples_per_second": 206.545, | |
| "eval_qasc_pairs_steps_per_second": 3.227, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_openbookqa_pairs_loss": 0.6995278000831604, | |
| "eval_openbookqa_pairs_runtime": 0.575, | |
| "eval_openbookqa_pairs_samples_per_second": 222.624, | |
| "eval_openbookqa_pairs_steps_per_second": 3.478, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_msmarco_pairs_loss": 0.37089064717292786, | |
| "eval_msmarco_pairs_runtime": 1.2866, | |
| "eval_msmarco_pairs_samples_per_second": 99.49, | |
| "eval_msmarco_pairs_steps_per_second": 1.555, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_nq_pairs_loss": 0.39078566431999207, | |
| "eval_nq_pairs_runtime": 2.3928, | |
| "eval_nq_pairs_samples_per_second": 53.493, | |
| "eval_nq_pairs_steps_per_second": 0.836, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_trivia_pairs_loss": 0.9003691077232361, | |
| "eval_trivia_pairs_runtime": 4.479, | |
| "eval_trivia_pairs_samples_per_second": 28.578, | |
| "eval_trivia_pairs_steps_per_second": 0.447, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_gooaq_pairs_loss": 0.5496390461921692, | |
| "eval_gooaq_pairs_runtime": 0.8735, | |
| "eval_gooaq_pairs_samples_per_second": 146.532, | |
| "eval_gooaq_pairs_steps_per_second": 2.29, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_paws-pos_loss": 0.024540428072214127, | |
| "eval_paws-pos_runtime": 0.6836, | |
| "eval_paws-pos_samples_per_second": 187.246, | |
| "eval_paws-pos_steps_per_second": 2.926, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9269662921348315, | |
| "grad_norm": 3.270443916320801, | |
| "learning_rate": 3.7525961967101216e-05, | |
| "loss": 0.5395, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.9438202247191011, | |
| "grad_norm": 2.608531951904297, | |
| "learning_rate": 3.7351572532692915e-05, | |
| "loss": 0.6874, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.9606741573033708, | |
| "grad_norm": 2.9995415210723877, | |
| "learning_rate": 3.717183680920136e-05, | |
| "loss": 0.5614, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.9775280898876404, | |
| "grad_norm": 3.42486572265625, | |
| "learning_rate": 3.698683259393809e-05, | |
| "loss": 0.5812, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.9943820224719101, | |
| "grad_norm": 1.8073548078536987, | |
| "learning_rate": 3.679663996464331e-05, | |
| "loss": 0.427, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.0112359550561798, | |
| "grad_norm": 3.2385106086730957, | |
| "learning_rate": 3.660134124482482e-05, | |
| "loss": 0.4603, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.0280898876404494, | |
| "grad_norm": 2.8062069416046143, | |
| "learning_rate": 3.640102096812488e-05, | |
| "loss": 0.6493, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.0449438202247192, | |
| "grad_norm": 2.6892104148864746, | |
| "learning_rate": 3.619576584173041e-05, | |
| "loss": 0.6646, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "grad_norm": 2.6583099365234375, | |
| "learning_rate": 3.598566470884244e-05, | |
| "loss": 0.7239, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_NLI-v2_cosine_accuracy": 1.0, | |
| "eval_NLI-v2_dot_accuracy": 0.0, | |
| "eval_NLI-v2_euclidean_accuracy": 1.0, | |
| "eval_NLI-v2_manhattan_accuracy": 1.0, | |
| "eval_NLI-v2_max_accuracy": 1.0, | |
| "eval_VitaminC_cosine_accuracy": 0.58984375, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8360881209373474, | |
| "eval_VitaminC_cosine_ap": 0.5601848253252508, | |
| "eval_VitaminC_cosine_f1": 0.6559999999999999, | |
| "eval_VitaminC_cosine_f1_threshold": 0.26484909653663635, | |
| "eval_VitaminC_cosine_precision": 0.4880952380952381, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.58203125, | |
| "eval_VitaminC_dot_accuracy_threshold": 314.279052734375, | |
| "eval_VitaminC_dot_ap": 0.5513292673695236, | |
| "eval_VitaminC_dot_f1": 0.6558265582655827, | |
| "eval_VitaminC_dot_f1_threshold": 126.1304931640625, | |
| "eval_VitaminC_dot_precision": 0.491869918699187, | |
| "eval_VitaminC_dot_recall": 0.983739837398374, | |
| "eval_VitaminC_euclidean_accuracy": 0.578125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 15.01893424987793, | |
| "eval_VitaminC_euclidean_ap": 0.5549132214851141, | |
| "eval_VitaminC_euclidean_f1": 0.6577540106951871, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.76571273803711, | |
| "eval_VitaminC_euclidean_precision": 0.4900398406374502, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.57421875, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 244.02972412109375, | |
| "eval_VitaminC_manhattan_ap": 0.5562338006363409, | |
| "eval_VitaminC_manhattan_f1": 0.6577540106951871, | |
| "eval_VitaminC_manhattan_f1_threshold": 498.5762634277344, | |
| "eval_VitaminC_manhattan_precision": 0.4900398406374502, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.58984375, | |
| "eval_VitaminC_max_accuracy_threshold": 314.279052734375, | |
| "eval_VitaminC_max_ap": 0.5601848253252508, | |
| "eval_VitaminC_max_f1": 0.6577540106951871, | |
| "eval_VitaminC_max_f1_threshold": 498.5762634277344, | |
| "eval_VitaminC_max_precision": 0.491869918699187, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5601848253252508, | |
| "eval_sts-test_pearson_cosine": 0.854968805652805, | |
| "eval_sts-test_pearson_dot": 0.8534110565503882, | |
| "eval_sts-test_pearson_euclidean": 0.8853384519331917, | |
| "eval_sts-test_pearson_manhattan": 0.8864271118397893, | |
| "eval_sts-test_pearson_max": 0.8864271118397893, | |
| "eval_sts-test_spearman_cosine": 0.8956917253507228, | |
| "eval_sts-test_spearman_dot": 0.877726389450295, | |
| "eval_sts-test_spearman_euclidean": 0.8875533307992096, | |
| "eval_sts-test_spearman_manhattan": 0.890112288382125, | |
| "eval_sts-test_spearman_max": 0.8956917253507228, | |
| "eval_vitaminc-pairs_loss": 2.3751792907714844, | |
| "eval_vitaminc-pairs_runtime": 1.4524, | |
| "eval_vitaminc-pairs_samples_per_second": 74.358, | |
| "eval_vitaminc-pairs_steps_per_second": 1.377, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_negation-triplets_loss": 1.580323338508606, | |
| "eval_negation-triplets_runtime": 0.301, | |
| "eval_negation-triplets_samples_per_second": 212.649, | |
| "eval_negation-triplets_steps_per_second": 3.323, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_scitail-pairs-pos_loss": 0.10438331216573715, | |
| "eval_scitail-pairs-pos_runtime": 0.363, | |
| "eval_scitail-pairs-pos_samples_per_second": 148.774, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.755, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_xsum-pairs_loss": 0.06599828600883484, | |
| "eval_xsum-pairs_runtime": 3.1538, | |
| "eval_xsum-pairs_samples_per_second": 40.586, | |
| "eval_xsum-pairs_steps_per_second": 0.634, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_sciq_pairs_loss": 0.033071305602788925, | |
| "eval_sciq_pairs_runtime": 3.2631, | |
| "eval_sciq_pairs_samples_per_second": 39.227, | |
| "eval_sciq_pairs_steps_per_second": 0.613, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_qasc_pairs_loss": 0.10076003521680832, | |
| "eval_qasc_pairs_runtime": 0.6385, | |
| "eval_qasc_pairs_samples_per_second": 200.46, | |
| "eval_qasc_pairs_steps_per_second": 3.132, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_openbookqa_pairs_loss": 0.6888580918312073, | |
| "eval_openbookqa_pairs_runtime": 0.6067, | |
| "eval_openbookqa_pairs_samples_per_second": 210.99, | |
| "eval_openbookqa_pairs_steps_per_second": 3.297, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_msmarco_pairs_loss": 0.335863322019577, | |
| "eval_msmarco_pairs_runtime": 1.3206, | |
| "eval_msmarco_pairs_samples_per_second": 96.922, | |
| "eval_msmarco_pairs_steps_per_second": 1.514, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_nq_pairs_loss": 0.3690747916698456, | |
| "eval_nq_pairs_runtime": 2.3983, | |
| "eval_nq_pairs_samples_per_second": 53.371, | |
| "eval_nq_pairs_steps_per_second": 0.834, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_trivia_pairs_loss": 0.881881594657898, | |
| "eval_trivia_pairs_runtime": 4.4261, | |
| "eval_trivia_pairs_samples_per_second": 28.919, | |
| "eval_trivia_pairs_steps_per_second": 0.452, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_gooaq_pairs_loss": 0.5453925728797913, | |
| "eval_gooaq_pairs_runtime": 0.8775, | |
| "eval_gooaq_pairs_samples_per_second": 145.871, | |
| "eval_gooaq_pairs_steps_per_second": 2.279, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0617977528089888, | |
| "eval_paws-pos_loss": 0.024841103702783585, | |
| "eval_paws-pos_runtime": 0.6851, | |
| "eval_paws-pos_samples_per_second": 186.83, | |
| "eval_paws-pos_steps_per_second": 2.919, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0786516853932584, | |
| "grad_norm": 2.942894220352173, | |
| "learning_rate": 3.5770808510220957e-05, | |
| "loss": 0.7593, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.095505617977528, | |
| "grad_norm": 2.636993169784546, | |
| "learning_rate": 3.5551290244821855e-05, | |
| "loss": 0.6877, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.1123595505617978, | |
| "grad_norm": 2.7987968921661377, | |
| "learning_rate": 3.5327204929543e-05, | |
| "loss": 0.5482, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.1292134831460674, | |
| "grad_norm": 2.3775548934936523, | |
| "learning_rate": 3.509864955809687e-05, | |
| "loss": 0.6047, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.146067415730337, | |
| "grad_norm": 1.982017993927002, | |
| "learning_rate": 3.4865723059027493e-05, | |
| "loss": 0.4358, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.1629213483146068, | |
| "grad_norm": 1.5380574464797974, | |
| "learning_rate": 3.462852625288999e-05, | |
| "loss": 0.3343, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.1797752808988764, | |
| "grad_norm": 3.152486562728882, | |
| "learning_rate": 3.438716180861106e-05, | |
| "loss": 0.5624, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.196629213483146, | |
| "grad_norm": 2.410505533218384, | |
| "learning_rate": 3.414173419904956e-05, | |
| "loss": 0.4578, | |
| "step": 213 | |
| } | |
| ], | |
| "logging_steps": 3, | |
| "max_steps": 534, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 107, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 320, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |