| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.40082858622475404, | |
| "eval_steps": 97, | |
| "global_step": 774, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010357327809425169, | |
| "grad_norm": 160.78689575195312, | |
| "learning_rate": 3.529411764705882e-07, | |
| "loss": 10.2062, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.020714655618850338, | |
| "grad_norm": 303.3471374511719, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 7.9221, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.031071983428275506, | |
| "grad_norm": 81.0146484375, | |
| "learning_rate": 1.2235294117647059e-06, | |
| "loss": 5.9499, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.041429311237700675, | |
| "grad_norm": 88.74897003173828, | |
| "learning_rate": 1.6470588235294118e-06, | |
| "loss": 6.0555, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_Qnli-dev_cosine_accuracy": 0.62109375, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.9108127355575562, | |
| "eval_Qnli-dev_cosine_ap": 0.6197524033200674, | |
| "eval_Qnli-dev_cosine_f1": 0.6396396396396397, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.8376526832580566, | |
| "eval_Qnli-dev_cosine_precision": 0.49534883720930234, | |
| "eval_Qnli-dev_cosine_recall": 0.902542372881356, | |
| "eval_Qnli-dev_dot_accuracy": 0.58984375, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 865.2555541992188, | |
| "eval_Qnli-dev_dot_ap": 0.5567642852275692, | |
| "eval_Qnli-dev_dot_f1": 0.6363636363636364, | |
| "eval_Qnli-dev_dot_f1_threshold": 691.9456787109375, | |
| "eval_Qnli-dev_dot_precision": 0.47863247863247865, | |
| "eval_Qnli-dev_dot_recall": 0.9491525423728814, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.609375, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 13.323524475097656, | |
| "eval_Qnli-dev_euclidean_ap": 0.6115116478210071, | |
| "eval_Qnli-dev_euclidean_f1": 0.6449612403100775, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 16.585830688476562, | |
| "eval_Qnli-dev_euclidean_precision": 0.508557457212714, | |
| "eval_Qnli-dev_euclidean_recall": 0.8813559322033898, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.619140625, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 406.1038818359375, | |
| "eval_Qnli-dev_manhattan_ap": 0.609904024113499, | |
| "eval_Qnli-dev_manhattan_f1": 0.6494345718901454, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 484.52716064453125, | |
| "eval_Qnli-dev_manhattan_precision": 0.5248041775456919, | |
| "eval_Qnli-dev_manhattan_recall": 0.8516949152542372, | |
| "eval_Qnli-dev_max_accuracy": 0.62109375, | |
| "eval_Qnli-dev_max_accuracy_threshold": 865.2555541992188, | |
| "eval_Qnli-dev_max_ap": 0.6197524033200674, | |
| "eval_Qnli-dev_max_f1": 0.6494345718901454, | |
| "eval_Qnli-dev_max_f1_threshold": 691.9456787109375, | |
| "eval_Qnli-dev_max_precision": 0.5248041775456919, | |
| "eval_Qnli-dev_max_recall": 0.9491525423728814, | |
| "eval_allNLI-dev_cosine_accuracy": 0.67578125, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.9652533531188965, | |
| "eval_allNLI-dev_cosine_ap": 0.4282858392784667, | |
| "eval_allNLI-dev_cosine_f1": 0.515527950310559, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.798592746257782, | |
| "eval_allNLI-dev_cosine_precision": 0.3524416135881104, | |
| "eval_allNLI-dev_cosine_recall": 0.9595375722543352, | |
| "eval_allNLI-dev_dot_accuracy": 0.666015625, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 968.9529418945312, | |
| "eval_allNLI-dev_dot_ap": 0.36425260705842155, | |
| "eval_allNLI-dev_dot_f1": 0.5162287480680062, | |
| "eval_allNLI-dev_dot_f1_threshold": 686.5814208984375, | |
| "eval_allNLI-dev_dot_precision": 0.35232067510548526, | |
| "eval_allNLI-dev_dot_recall": 0.9653179190751445, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.67578125, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 8.16073226928711, | |
| "eval_allNLI-dev_euclidean_ap": 0.4333583117036793, | |
| "eval_allNLI-dev_euclidean_f1": 0.5164319248826291, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 18.877037048339844, | |
| "eval_allNLI-dev_euclidean_precision": 0.3540772532188841, | |
| "eval_allNLI-dev_euclidean_recall": 0.953757225433526, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.67578125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 226.18099975585938, | |
| "eval_allNLI-dev_manhattan_ap": 0.4400955405569059, | |
| "eval_allNLI-dev_manhattan_f1": 0.5179407176287052, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 570.2012329101562, | |
| "eval_allNLI-dev_manhattan_precision": 0.3547008547008547, | |
| "eval_allNLI-dev_manhattan_recall": 0.9595375722543352, | |
| "eval_allNLI-dev_max_accuracy": 0.67578125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 968.9529418945312, | |
| "eval_allNLI-dev_max_ap": 0.4400955405569059, | |
| "eval_allNLI-dev_max_f1": 0.5179407176287052, | |
| "eval_allNLI-dev_max_f1_threshold": 686.5814208984375, | |
| "eval_allNLI-dev_max_precision": 0.3547008547008547, | |
| "eval_allNLI-dev_max_recall": 0.9653179190751445, | |
| "eval_sequential_score": 0.6197524033200674, | |
| "eval_sts-test_pearson_cosine": 0.6170839897033953, | |
| "eval_sts-test_pearson_dot": 0.43346770865150264, | |
| "eval_sts-test_pearson_euclidean": 0.6474775644966124, | |
| "eval_sts-test_pearson_manhattan": 0.6616828287248389, | |
| "eval_sts-test_pearson_max": 0.6616828287248389, | |
| "eval_sts-test_spearman_cosine": 0.6552392427969004, | |
| "eval_sts-test_spearman_dot": 0.4585595522909849, | |
| "eval_sts-test_spearman_euclidean": 0.652406174691995, | |
| "eval_sts-test_spearman_manhattan": 0.6662387448368152, | |
| "eval_sts-test_spearman_max": 0.6662387448368152, | |
| "eval_vitaminc-pairs_loss": 3.7554073333740234, | |
| "eval_vitaminc-pairs_runtime": 4.7418, | |
| "eval_vitaminc-pairs_samples_per_second": 26.994, | |
| "eval_vitaminc-pairs_steps_per_second": 0.211, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_negation-triplets_loss": 3.6897997856140137, | |
| "eval_negation-triplets_runtime": 3.1578, | |
| "eval_negation-triplets_samples_per_second": 40.534, | |
| "eval_negation-triplets_steps_per_second": 0.317, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_scitail-pairs-pos_loss": 0.45563364028930664, | |
| "eval_scitail-pairs-pos_runtime": 2.7223, | |
| "eval_scitail-pairs-pos_samples_per_second": 47.02, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.367, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_scitail-pairs-qa_loss": 1.115855097770691, | |
| "eval_scitail-pairs-qa_runtime": 2.2597, | |
| "eval_scitail-pairs-qa_samples_per_second": 56.646, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.443, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_xsum-pairs_loss": 3.2118453979492188, | |
| "eval_xsum-pairs_runtime": 3.0538, | |
| "eval_xsum-pairs_samples_per_second": 41.915, | |
| "eval_xsum-pairs_steps_per_second": 0.327, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_sciq_pairs_loss": 0.484823614358902, | |
| "eval_sciq_pairs_runtime": 3.959, | |
| "eval_sciq_pairs_samples_per_second": 32.331, | |
| "eval_sciq_pairs_steps_per_second": 0.253, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_qasc_pairs_loss": 2.8566131591796875, | |
| "eval_qasc_pairs_runtime": 2.1087, | |
| "eval_qasc_pairs_samples_per_second": 60.701, | |
| "eval_qasc_pairs_steps_per_second": 0.474, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_openbookqa_pairs_loss": 2.1501104831695557, | |
| "eval_openbookqa_pairs_runtime": 2.2555, | |
| "eval_openbookqa_pairs_samples_per_second": 56.751, | |
| "eval_openbookqa_pairs_steps_per_second": 0.443, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_msmarco_pairs_loss": 4.395960807800293, | |
| "eval_msmarco_pairs_runtime": 2.2407, | |
| "eval_msmarco_pairs_samples_per_second": 57.125, | |
| "eval_msmarco_pairs_steps_per_second": 0.446, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_nq_pairs_loss": 4.488173484802246, | |
| "eval_nq_pairs_runtime": 2.7484, | |
| "eval_nq_pairs_samples_per_second": 46.572, | |
| "eval_nq_pairs_steps_per_second": 0.364, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_trivia_pairs_loss": 4.023955345153809, | |
| "eval_trivia_pairs_runtime": 3.7908, | |
| "eval_trivia_pairs_samples_per_second": 33.766, | |
| "eval_trivia_pairs_steps_per_second": 0.264, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_gooaq_pairs_loss": 3.383638858795166, | |
| "eval_gooaq_pairs_runtime": 2.1349, | |
| "eval_gooaq_pairs_samples_per_second": 59.957, | |
| "eval_gooaq_pairs_steps_per_second": 0.468, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_paws-pos_loss": 0.12275903671979904, | |
| "eval_paws-pos_runtime": 2.2818, | |
| "eval_paws-pos_samples_per_second": 56.095, | |
| "eval_paws-pos_steps_per_second": 0.438, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.050233039875712066, | |
| "eval_global_dataset_loss": 1.9564138650894165, | |
| "eval_global_dataset_runtime": 10.2217, | |
| "eval_global_dataset_samples_per_second": 40.698, | |
| "eval_global_dataset_steps_per_second": 0.391, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05178663904712584, | |
| "grad_norm": 83.8360824584961, | |
| "learning_rate": 2.1176470588235296e-06, | |
| "loss": 4.0315, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06214396685655101, | |
| "grad_norm": 325.5680236816406, | |
| "learning_rate": 2.588235294117647e-06, | |
| "loss": 1.6348, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07250129466597618, | |
| "grad_norm": 106.99758911132812, | |
| "learning_rate": 3.0588235294117647e-06, | |
| "loss": 1.1866, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08285862247540135, | |
| "grad_norm": 30.390771865844727, | |
| "learning_rate": 3.5294117647058825e-06, | |
| "loss": 0.6138, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09321595028482652, | |
| "grad_norm": 39.691532135009766, | |
| "learning_rate": 4e-06, | |
| "loss": 0.5244, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_Qnli-dev_cosine_accuracy": 0.666015625, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7696025967597961, | |
| "eval_Qnli-dev_cosine_ap": 0.693851901846308, | |
| "eval_Qnli-dev_cosine_f1": 0.6625000000000001, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6638460159301758, | |
| "eval_Qnli-dev_cosine_precision": 0.5247524752475248, | |
| "eval_Qnli-dev_cosine_recall": 0.8983050847457628, | |
| "eval_Qnli-dev_dot_accuracy": 0.6796875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 822.6981201171875, | |
| "eval_Qnli-dev_dot_ap": 0.6603086879421342, | |
| "eval_Qnli-dev_dot_f1": 0.657856093979442, | |
| "eval_Qnli-dev_dot_f1_threshold": 618.4547119140625, | |
| "eval_Qnli-dev_dot_precision": 0.503370786516854, | |
| "eval_Qnli-dev_dot_recall": 0.9491525423728814, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.666015625, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 19.874027252197266, | |
| "eval_Qnli-dev_euclidean_ap": 0.694158709095853, | |
| "eval_Qnli-dev_euclidean_f1": 0.6630236794171221, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 23.005264282226562, | |
| "eval_Qnli-dev_euclidean_precision": 0.5814696485623003, | |
| "eval_Qnli-dev_euclidean_recall": 0.7711864406779662, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.66796875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 624.285888671875, | |
| "eval_Qnli-dev_manhattan_ap": 0.692633721446368, | |
| "eval_Qnli-dev_manhattan_f1": 0.6641366223908918, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 696.709716796875, | |
| "eval_Qnli-dev_manhattan_precision": 0.6013745704467354, | |
| "eval_Qnli-dev_manhattan_recall": 0.7415254237288136, | |
| "eval_Qnli-dev_max_accuracy": 0.6796875, | |
| "eval_Qnli-dev_max_accuracy_threshold": 822.6981201171875, | |
| "eval_Qnli-dev_max_ap": 0.694158709095853, | |
| "eval_Qnli-dev_max_f1": 0.6641366223908918, | |
| "eval_Qnli-dev_max_f1_threshold": 696.709716796875, | |
| "eval_Qnli-dev_max_precision": 0.6013745704467354, | |
| "eval_Qnli-dev_max_recall": 0.9491525423728814, | |
| "eval_allNLI-dev_cosine_accuracy": 0.701171875, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.854247510433197, | |
| "eval_allNLI-dev_cosine_ap": 0.5504250327111149, | |
| "eval_allNLI-dev_cosine_f1": 0.567287784679089, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7080726623535156, | |
| "eval_allNLI-dev_cosine_precision": 0.44193548387096776, | |
| "eval_allNLI-dev_cosine_recall": 0.791907514450867, | |
| "eval_allNLI-dev_dot_accuracy": 0.69921875, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 885.8963623046875, | |
| "eval_allNLI-dev_dot_ap": 0.5371398846089106, | |
| "eval_allNLI-dev_dot_f1": 0.5720338983050848, | |
| "eval_allNLI-dev_dot_f1_threshold": 732.1597290039062, | |
| "eval_allNLI-dev_dot_precision": 0.451505016722408, | |
| "eval_allNLI-dev_dot_recall": 0.7803468208092486, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.701171875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 16.9801082611084, | |
| "eval_allNLI-dev_euclidean_ap": 0.5503780840587245, | |
| "eval_allNLI-dev_euclidean_f1": 0.5671641791044777, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 24.19074821472168, | |
| "eval_allNLI-dev_euclidean_precision": 0.44932432432432434, | |
| "eval_allNLI-dev_euclidean_recall": 0.7687861271676301, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.703125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 529.9462280273438, | |
| "eval_allNLI-dev_manhattan_ap": 0.5524969745859143, | |
| "eval_allNLI-dev_manhattan_f1": 0.5638297872340425, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 826.8560791015625, | |
| "eval_allNLI-dev_manhattan_precision": 0.40664961636828645, | |
| "eval_allNLI-dev_manhattan_recall": 0.9190751445086706, | |
| "eval_allNLI-dev_max_accuracy": 0.703125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 885.8963623046875, | |
| "eval_allNLI-dev_max_ap": 0.5524969745859143, | |
| "eval_allNLI-dev_max_f1": 0.5720338983050848, | |
| "eval_allNLI-dev_max_f1_threshold": 826.8560791015625, | |
| "eval_allNLI-dev_max_precision": 0.451505016722408, | |
| "eval_allNLI-dev_max_recall": 0.9190751445086706, | |
| "eval_sequential_score": 0.694158709095853, | |
| "eval_sts-test_pearson_cosine": 0.8866994033223972, | |
| "eval_sts-test_pearson_dot": 0.8712266973511624, | |
| "eval_sts-test_pearson_euclidean": 0.9028053322103908, | |
| "eval_sts-test_pearson_manhattan": 0.9029714248344419, | |
| "eval_sts-test_pearson_max": 0.9029714248344419, | |
| "eval_sts-test_spearman_cosine": 0.8941879764786184, | |
| "eval_sts-test_spearman_dot": 0.8632849034222648, | |
| "eval_sts-test_spearman_euclidean": 0.8944520984233506, | |
| "eval_sts-test_spearman_manhattan": 0.8945218656398598, | |
| "eval_sts-test_spearman_max": 0.8945218656398598, | |
| "eval_vitaminc-pairs_loss": 3.507073163986206, | |
| "eval_vitaminc-pairs_runtime": 4.4774, | |
| "eval_vitaminc-pairs_samples_per_second": 28.588, | |
| "eval_vitaminc-pairs_steps_per_second": 0.223, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_negation-triplets_loss": 1.1223009824752808, | |
| "eval_negation-triplets_runtime": 3.102, | |
| "eval_negation-triplets_samples_per_second": 41.264, | |
| "eval_negation-triplets_steps_per_second": 0.322, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_scitail-pairs-pos_loss": 0.06560208648443222, | |
| "eval_scitail-pairs-pos_runtime": 2.6151, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.946, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.382, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_scitail-pairs-qa_loss": 0.044671397656202316, | |
| "eval_scitail-pairs-qa_runtime": 2.2115, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.879, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.452, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_xsum-pairs_loss": 0.07691845297813416, | |
| "eval_xsum-pairs_runtime": 3.043, | |
| "eval_xsum-pairs_samples_per_second": 42.064, | |
| "eval_xsum-pairs_steps_per_second": 0.329, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_sciq_pairs_loss": 0.12039273232221603, | |
| "eval_sciq_pairs_runtime": 3.878, | |
| "eval_sciq_pairs_samples_per_second": 33.007, | |
| "eval_sciq_pairs_steps_per_second": 0.258, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_qasc_pairs_loss": 0.36198654770851135, | |
| "eval_qasc_pairs_runtime": 2.0543, | |
| "eval_qasc_pairs_samples_per_second": 62.307, | |
| "eval_qasc_pairs_steps_per_second": 0.487, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_openbookqa_pairs_loss": 0.5711529850959778, | |
| "eval_openbookqa_pairs_runtime": 2.2213, | |
| "eval_openbookqa_pairs_samples_per_second": 57.624, | |
| "eval_openbookqa_pairs_steps_per_second": 0.45, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_msmarco_pairs_loss": 0.3250836133956909, | |
| "eval_msmarco_pairs_runtime": 2.22, | |
| "eval_msmarco_pairs_samples_per_second": 57.657, | |
| "eval_msmarco_pairs_steps_per_second": 0.45, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_nq_pairs_loss": 0.4249531030654907, | |
| "eval_nq_pairs_runtime": 2.7189, | |
| "eval_nq_pairs_samples_per_second": 47.079, | |
| "eval_nq_pairs_steps_per_second": 0.368, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_trivia_pairs_loss": 0.2965388894081116, | |
| "eval_trivia_pairs_runtime": 3.7556, | |
| "eval_trivia_pairs_samples_per_second": 34.082, | |
| "eval_trivia_pairs_steps_per_second": 0.266, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_gooaq_pairs_loss": 0.2151084989309311, | |
| "eval_gooaq_pairs_runtime": 2.1122, | |
| "eval_gooaq_pairs_samples_per_second": 60.601, | |
| "eval_gooaq_pairs_steps_per_second": 0.473, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_paws-pos_loss": 0.0295370165258646, | |
| "eval_paws-pos_runtime": 2.2123, | |
| "eval_paws-pos_samples_per_second": 57.86, | |
| "eval_paws-pos_steps_per_second": 0.452, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10046607975142413, | |
| "eval_global_dataset_loss": 0.35498398542404175, | |
| "eval_global_dataset_runtime": 10.1407, | |
| "eval_global_dataset_samples_per_second": 41.023, | |
| "eval_global_dataset_steps_per_second": 0.394, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10357327809425168, | |
| "grad_norm": 43.11693572998047, | |
| "learning_rate": 4.470588235294118e-06, | |
| "loss": 0.376, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11393060590367685, | |
| "grad_norm": 31.125375747680664, | |
| "learning_rate": 4.941176470588235e-06, | |
| "loss": 0.2782, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12428793371310203, | |
| "grad_norm": 14.048110961914062, | |
| "learning_rate": 5.411764705882353e-06, | |
| "loss": 0.2391, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.13464526152252718, | |
| "grad_norm": 5.956579685211182, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 0.2767, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.14500258933195237, | |
| "grad_norm": 14.470146179199219, | |
| "learning_rate": 6.352941176470589e-06, | |
| "loss": 0.2359, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_Qnli-dev_cosine_accuracy": 0.6875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7567152976989746, | |
| "eval_Qnli-dev_cosine_ap": 0.7133123361631746, | |
| "eval_Qnli-dev_cosine_f1": 0.6853146853146853, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6536699533462524, | |
| "eval_Qnli-dev_cosine_precision": 0.5833333333333334, | |
| "eval_Qnli-dev_cosine_recall": 0.8305084745762712, | |
| "eval_Qnli-dev_dot_accuracy": 0.673828125, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 731.5150756835938, | |
| "eval_Qnli-dev_dot_ap": 0.6890325242500185, | |
| "eval_Qnli-dev_dot_f1": 0.6782006920415226, | |
| "eval_Qnli-dev_dot_f1_threshold": 621.156982421875, | |
| "eval_Qnli-dev_dot_precision": 0.5730994152046783, | |
| "eval_Qnli-dev_dot_recall": 0.8305084745762712, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.6875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 21.166996002197266, | |
| "eval_Qnli-dev_euclidean_ap": 0.717782618584373, | |
| "eval_Qnli-dev_euclidean_f1": 0.6832740213523131, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 25.534191131591797, | |
| "eval_Qnli-dev_euclidean_precision": 0.588957055214724, | |
| "eval_Qnli-dev_euclidean_recall": 0.8135593220338984, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.689453125, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 717.0855712890625, | |
| "eval_Qnli-dev_manhattan_ap": 0.7178394918687495, | |
| "eval_Qnli-dev_manhattan_f1": 0.6815068493150686, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 809.9966430664062, | |
| "eval_Qnli-dev_manhattan_precision": 0.5718390804597702, | |
| "eval_Qnli-dev_manhattan_recall": 0.8432203389830508, | |
| "eval_Qnli-dev_max_accuracy": 0.689453125, | |
| "eval_Qnli-dev_max_accuracy_threshold": 731.5150756835938, | |
| "eval_Qnli-dev_max_ap": 0.7178394918687495, | |
| "eval_Qnli-dev_max_f1": 0.6853146853146853, | |
| "eval_Qnli-dev_max_f1_threshold": 809.9966430664062, | |
| "eval_Qnli-dev_max_precision": 0.588957055214724, | |
| "eval_Qnli-dev_max_recall": 0.8432203389830508, | |
| "eval_allNLI-dev_cosine_accuracy": 0.71484375, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8485724329948425, | |
| "eval_allNLI-dev_cosine_ap": 0.5777522094864251, | |
| "eval_allNLI-dev_cosine_f1": 0.5925925925925926, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7124052047729492, | |
| "eval_allNLI-dev_cosine_precision": 0.4942084942084942, | |
| "eval_allNLI-dev_cosine_recall": 0.7398843930635838, | |
| "eval_allNLI-dev_dot_accuracy": 0.71484375, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 835.6192016601562, | |
| "eval_allNLI-dev_dot_ap": 0.5708546535940942, | |
| "eval_allNLI-dev_dot_f1": 0.5931372549019609, | |
| "eval_allNLI-dev_dot_f1_threshold": 712.94482421875, | |
| "eval_allNLI-dev_dot_precision": 0.5148936170212766, | |
| "eval_allNLI-dev_dot_recall": 0.6994219653179191, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.712890625, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 15.772256851196289, | |
| "eval_allNLI-dev_euclidean_ap": 0.5773033114664347, | |
| "eval_allNLI-dev_euclidean_f1": 0.5957446808510639, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 24.513042449951172, | |
| "eval_allNLI-dev_euclidean_precision": 0.4713804713804714, | |
| "eval_allNLI-dev_euclidean_recall": 0.8092485549132948, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.71484375, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 494.4720153808594, | |
| "eval_allNLI-dev_manhattan_ap": 0.5787277750430182, | |
| "eval_allNLI-dev_manhattan_f1": 0.597457627118644, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 764.1075439453125, | |
| "eval_allNLI-dev_manhattan_precision": 0.47157190635451507, | |
| "eval_allNLI-dev_manhattan_recall": 0.815028901734104, | |
| "eval_allNLI-dev_max_accuracy": 0.71484375, | |
| "eval_allNLI-dev_max_accuracy_threshold": 835.6192016601562, | |
| "eval_allNLI-dev_max_ap": 0.5787277750430182, | |
| "eval_allNLI-dev_max_f1": 0.597457627118644, | |
| "eval_allNLI-dev_max_f1_threshold": 764.1075439453125, | |
| "eval_allNLI-dev_max_precision": 0.5148936170212766, | |
| "eval_allNLI-dev_max_recall": 0.815028901734104, | |
| "eval_sequential_score": 0.7178394918687495, | |
| "eval_sts-test_pearson_cosine": 0.9080888281681364, | |
| "eval_sts-test_pearson_dot": 0.8993720999648187, | |
| "eval_sts-test_pearson_euclidean": 0.9185021221297063, | |
| "eval_sts-test_pearson_manhattan": 0.9182084064307341, | |
| "eval_sts-test_pearson_max": 0.9185021221297063, | |
| "eval_sts-test_spearman_cosine": 0.9145502926755805, | |
| "eval_sts-test_spearman_dot": 0.8990795555767088, | |
| "eval_sts-test_spearman_euclidean": 0.9143005806370166, | |
| "eval_sts-test_spearman_manhattan": 0.9141107457861942, | |
| "eval_sts-test_spearman_max": 0.9145502926755805, | |
| "eval_vitaminc-pairs_loss": 3.4645299911499023, | |
| "eval_vitaminc-pairs_runtime": 4.4497, | |
| "eval_vitaminc-pairs_samples_per_second": 28.766, | |
| "eval_vitaminc-pairs_steps_per_second": 0.225, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_negation-triplets_loss": 0.8774887323379517, | |
| "eval_negation-triplets_runtime": 3.1401, | |
| "eval_negation-triplets_samples_per_second": 40.764, | |
| "eval_negation-triplets_steps_per_second": 0.318, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_scitail-pairs-pos_loss": 0.029673559591174126, | |
| "eval_scitail-pairs-pos_runtime": 2.6642, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.044, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.375, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_scitail-pairs-qa_loss": 0.011800204403698444, | |
| "eval_scitail-pairs-qa_runtime": 2.1861, | |
| "eval_scitail-pairs-qa_samples_per_second": 58.551, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.457, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_xsum-pairs_loss": 0.017930012196302414, | |
| "eval_xsum-pairs_runtime": 3.0255, | |
| "eval_xsum-pairs_samples_per_second": 42.307, | |
| "eval_xsum-pairs_steps_per_second": 0.331, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_sciq_pairs_loss": 0.09765021502971649, | |
| "eval_sciq_pairs_runtime": 3.8726, | |
| "eval_sciq_pairs_samples_per_second": 33.053, | |
| "eval_sciq_pairs_steps_per_second": 0.258, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_qasc_pairs_loss": 0.3064229488372803, | |
| "eval_qasc_pairs_runtime": 2.1307, | |
| "eval_qasc_pairs_samples_per_second": 60.075, | |
| "eval_qasc_pairs_steps_per_second": 0.469, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_openbookqa_pairs_loss": 0.46111759543418884, | |
| "eval_openbookqa_pairs_runtime": 2.2685, | |
| "eval_openbookqa_pairs_samples_per_second": 56.424, | |
| "eval_openbookqa_pairs_steps_per_second": 0.441, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_msmarco_pairs_loss": 0.08168309926986694, | |
| "eval_msmarco_pairs_runtime": 2.2657, | |
| "eval_msmarco_pairs_samples_per_second": 56.495, | |
| "eval_msmarco_pairs_steps_per_second": 0.441, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_nq_pairs_loss": 0.13220462203025818, | |
| "eval_nq_pairs_runtime": 2.7139, | |
| "eval_nq_pairs_samples_per_second": 47.164, | |
| "eval_nq_pairs_steps_per_second": 0.368, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_trivia_pairs_loss": 0.1532345414161682, | |
| "eval_trivia_pairs_runtime": 3.76, | |
| "eval_trivia_pairs_samples_per_second": 34.043, | |
| "eval_trivia_pairs_steps_per_second": 0.266, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_gooaq_pairs_loss": 0.10126010328531265, | |
| "eval_gooaq_pairs_runtime": 2.1372, | |
| "eval_gooaq_pairs_samples_per_second": 59.892, | |
| "eval_gooaq_pairs_steps_per_second": 0.468, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_paws-pos_loss": 0.021147189661860466, | |
| "eval_paws-pos_runtime": 2.2138, | |
| "eval_paws-pos_samples_per_second": 57.819, | |
| "eval_paws-pos_steps_per_second": 0.452, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1506991196271362, | |
| "eval_global_dataset_loss": 0.2509276270866394, | |
| "eval_global_dataset_runtime": 10.154, | |
| "eval_global_dataset_samples_per_second": 40.969, | |
| "eval_global_dataset_steps_per_second": 0.394, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.15535991714137753, | |
| "grad_norm": 6.319842338562012, | |
| "learning_rate": 6.823529411764706e-06, | |
| "loss": 0.1505, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1657172449508027, | |
| "grad_norm": 16.11246109008789, | |
| "learning_rate": 7.294117647058823e-06, | |
| "loss": 0.1473, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.17607457276022787, | |
| "grad_norm": 9.587472915649414, | |
| "learning_rate": 7.764705882352943e-06, | |
| "loss": 0.1614, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.18643190056965303, | |
| "grad_norm": 21.551036834716797, | |
| "learning_rate": 8.23529411764706e-06, | |
| "loss": 0.1834, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.1967892283790782, | |
| "grad_norm": 7.1668548583984375, | |
| "learning_rate": 8.705882352941177e-06, | |
| "loss": 0.164, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_Qnli-dev_cosine_accuracy": 0.685546875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.6639062166213989, | |
| "eval_Qnli-dev_cosine_ap": 0.7263308094103806, | |
| "eval_Qnli-dev_cosine_f1": 0.6986754966887417, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.576126754283905, | |
| "eval_Qnli-dev_cosine_precision": 0.5733695652173914, | |
| "eval_Qnli-dev_cosine_recall": 0.8940677966101694, | |
| "eval_Qnli-dev_dot_accuracy": 0.677734375, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 591.922607421875, | |
| "eval_Qnli-dev_dot_ap": 0.7011149935901715, | |
| "eval_Qnli-dev_dot_f1": 0.6837606837606838, | |
| "eval_Qnli-dev_dot_f1_threshold": 505.88946533203125, | |
| "eval_Qnli-dev_dot_precision": 0.5730659025787965, | |
| "eval_Qnli-dev_dot_recall": 0.847457627118644, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.693359375, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 23.223190307617188, | |
| "eval_Qnli-dev_euclidean_ap": 0.7305021849937567, | |
| "eval_Qnli-dev_euclidean_f1": 0.698205546492659, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 27.346588134765625, | |
| "eval_Qnli-dev_euclidean_precision": 0.5676392572944297, | |
| "eval_Qnli-dev_euclidean_recall": 0.9067796610169492, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.6953125, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 733.809814453125, | |
| "eval_Qnli-dev_manhattan_ap": 0.7286452491858623, | |
| "eval_Qnli-dev_manhattan_f1": 0.7008264462809918, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 845.1378784179688, | |
| "eval_Qnli-dev_manhattan_precision": 0.5745257452574526, | |
| "eval_Qnli-dev_manhattan_recall": 0.8983050847457628, | |
| "eval_Qnli-dev_max_accuracy": 0.6953125, | |
| "eval_Qnli-dev_max_accuracy_threshold": 733.809814453125, | |
| "eval_Qnli-dev_max_ap": 0.7305021849937567, | |
| "eval_Qnli-dev_max_f1": 0.7008264462809918, | |
| "eval_Qnli-dev_max_f1_threshold": 845.1378784179688, | |
| "eval_Qnli-dev_max_precision": 0.5745257452574526, | |
| "eval_Qnli-dev_max_recall": 0.9067796610169492, | |
| "eval_allNLI-dev_cosine_accuracy": 0.720703125, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8122999668121338, | |
| "eval_allNLI-dev_cosine_ap": 0.6057945129739214, | |
| "eval_allNLI-dev_cosine_f1": 0.6099585062240664, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6289657950401306, | |
| "eval_allNLI-dev_cosine_precision": 0.47572815533980584, | |
| "eval_allNLI-dev_cosine_recall": 0.8497109826589595, | |
| "eval_allNLI-dev_dot_accuracy": 0.71875, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 745.8334350585938, | |
| "eval_allNLI-dev_dot_ap": 0.5916353965674287, | |
| "eval_allNLI-dev_dot_f1": 0.610655737704918, | |
| "eval_allNLI-dev_dot_f1_threshold": 540.4627075195312, | |
| "eval_allNLI-dev_dot_precision": 0.473015873015873, | |
| "eval_allNLI-dev_dot_recall": 0.861271676300578, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.71875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 18.420812606811523, | |
| "eval_allNLI-dev_euclidean_ap": 0.6036307863078971, | |
| "eval_allNLI-dev_euclidean_f1": 0.6182572614107884, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 25.34260368347168, | |
| "eval_allNLI-dev_euclidean_precision": 0.48220064724919093, | |
| "eval_allNLI-dev_euclidean_recall": 0.861271676300578, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.72265625, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 600.5528564453125, | |
| "eval_allNLI-dev_manhattan_ap": 0.60455800678133, | |
| "eval_allNLI-dev_manhattan_f1": 0.6170212765957447, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 781.2642822265625, | |
| "eval_allNLI-dev_manhattan_precision": 0.4882154882154882, | |
| "eval_allNLI-dev_manhattan_recall": 0.838150289017341, | |
| "eval_allNLI-dev_max_accuracy": 0.72265625, | |
| "eval_allNLI-dev_max_accuracy_threshold": 745.8334350585938, | |
| "eval_allNLI-dev_max_ap": 0.6057945129739214, | |
| "eval_allNLI-dev_max_f1": 0.6182572614107884, | |
| "eval_allNLI-dev_max_f1_threshold": 781.2642822265625, | |
| "eval_allNLI-dev_max_precision": 0.4882154882154882, | |
| "eval_allNLI-dev_max_recall": 0.861271676300578, | |
| "eval_sequential_score": 0.7305021849937567, | |
| "eval_sts-test_pearson_cosine": 0.9127094359947289, | |
| "eval_sts-test_pearson_dot": 0.895861018162462, | |
| "eval_sts-test_pearson_euclidean": 0.9199643208978237, | |
| "eval_sts-test_pearson_manhattan": 0.9196227536115376, | |
| "eval_sts-test_pearson_max": 0.9199643208978237, | |
| "eval_sts-test_spearman_cosine": 0.9147098309224408, | |
| "eval_sts-test_spearman_dot": 0.8908736085574486, | |
| "eval_sts-test_spearman_euclidean": 0.9157628281029806, | |
| "eval_sts-test_spearman_manhattan": 0.9154038334840987, | |
| "eval_sts-test_spearman_max": 0.9157628281029806, | |
| "eval_vitaminc-pairs_loss": 2.9898574352264404, | |
| "eval_vitaminc-pairs_runtime": 4.5557, | |
| "eval_vitaminc-pairs_samples_per_second": 28.097, | |
| "eval_vitaminc-pairs_steps_per_second": 0.22, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_negation-triplets_loss": 0.761246919631958, | |
| "eval_negation-triplets_runtime": 3.3164, | |
| "eval_negation-triplets_samples_per_second": 38.596, | |
| "eval_negation-triplets_steps_per_second": 0.302, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_scitail-pairs-pos_loss": 0.0324205681681633, | |
| "eval_scitail-pairs-pos_runtime": 2.7411, | |
| "eval_scitail-pairs-pos_samples_per_second": 46.696, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.365, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_scitail-pairs-qa_loss": 0.0026867901906371117, | |
| "eval_scitail-pairs-qa_runtime": 2.4223, | |
| "eval_scitail-pairs-qa_samples_per_second": 52.843, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.413, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_xsum-pairs_loss": 0.00942266546189785, | |
| "eval_xsum-pairs_runtime": 3.1951, | |
| "eval_xsum-pairs_samples_per_second": 40.061, | |
| "eval_xsum-pairs_steps_per_second": 0.313, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_sciq_pairs_loss": 0.09475678950548172, | |
| "eval_sciq_pairs_runtime": 4.1048, | |
| "eval_sciq_pairs_samples_per_second": 31.183, | |
| "eval_sciq_pairs_steps_per_second": 0.244, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_qasc_pairs_loss": 0.2342282086610794, | |
| "eval_qasc_pairs_runtime": 2.27, | |
| "eval_qasc_pairs_samples_per_second": 56.388, | |
| "eval_qasc_pairs_steps_per_second": 0.441, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_openbookqa_pairs_loss": 0.4459604024887085, | |
| "eval_openbookqa_pairs_runtime": 2.3513, | |
| "eval_openbookqa_pairs_samples_per_second": 54.438, | |
| "eval_openbookqa_pairs_steps_per_second": 0.425, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_msmarco_pairs_loss": 0.2047792673110962, | |
| "eval_msmarco_pairs_runtime": 2.3749, | |
| "eval_msmarco_pairs_samples_per_second": 53.897, | |
| "eval_msmarco_pairs_steps_per_second": 0.421, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_nq_pairs_loss": 0.12687399983406067, | |
| "eval_nq_pairs_runtime": 2.8216, | |
| "eval_nq_pairs_samples_per_second": 45.365, | |
| "eval_nq_pairs_steps_per_second": 0.354, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_trivia_pairs_loss": 0.208355113863945, | |
| "eval_trivia_pairs_runtime": 3.8421, | |
| "eval_trivia_pairs_samples_per_second": 33.315, | |
| "eval_trivia_pairs_steps_per_second": 0.26, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_gooaq_pairs_loss": 0.10170701891183853, | |
| "eval_gooaq_pairs_runtime": 2.3264, | |
| "eval_gooaq_pairs_samples_per_second": 55.02, | |
| "eval_gooaq_pairs_steps_per_second": 0.43, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_paws-pos_loss": 0.0226531233638525, | |
| "eval_paws-pos_runtime": 2.4028, | |
| "eval_paws-pos_samples_per_second": 53.271, | |
| "eval_paws-pos_steps_per_second": 0.416, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20093215950284826, | |
| "eval_global_dataset_loss": 0.22992311418056488, | |
| "eval_global_dataset_runtime": 10.4483, | |
| "eval_global_dataset_samples_per_second": 39.815, | |
| "eval_global_dataset_steps_per_second": 0.383, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.20714655618850336, | |
| "grad_norm": 0.033974967896938324, | |
| "learning_rate": 9.176470588235295e-06, | |
| "loss": 0.1426, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21750388399792853, | |
| "grad_norm": 8.04489517211914, | |
| "learning_rate": 9.647058823529412e-06, | |
| "loss": 0.1838, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2278612118073537, | |
| "grad_norm": 37.961544036865234, | |
| "learning_rate": 1.0117647058823531e-05, | |
| "loss": 0.1324, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.23821853961677888, | |
| "grad_norm": 9.86117172241211, | |
| "learning_rate": 1.0588235294117648e-05, | |
| "loss": 0.1242, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.24857586742620405, | |
| "grad_norm": 15.386984825134277, | |
| "learning_rate": 1.1058823529411766e-05, | |
| "loss": 0.2166, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_Qnli-dev_cosine_accuracy": 0.70703125, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.6728878021240234, | |
| "eval_Qnli-dev_cosine_ap": 0.7490927864840249, | |
| "eval_Qnli-dev_cosine_f1": 0.7092198581560283, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6182924509048462, | |
| "eval_Qnli-dev_cosine_precision": 0.6097560975609756, | |
| "eval_Qnli-dev_cosine_recall": 0.847457627118644, | |
| "eval_Qnli-dev_dot_accuracy": 0.67578125, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 664.639404296875, | |
| "eval_Qnli-dev_dot_ap": 0.7143774472576185, | |
| "eval_Qnli-dev_dot_f1": 0.7084019769357496, | |
| "eval_Qnli-dev_dot_f1_threshold": 506.4283447265625, | |
| "eval_Qnli-dev_dot_precision": 0.5795148247978437, | |
| "eval_Qnli-dev_dot_recall": 0.9110169491525424, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.7109375, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 23.347135543823242, | |
| "eval_Qnli-dev_euclidean_ap": 0.7578270539486094, | |
| "eval_Qnli-dev_euclidean_f1": 0.712041884816754, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 26.101980209350586, | |
| "eval_Qnli-dev_euclidean_precision": 0.6053412462908012, | |
| "eval_Qnli-dev_euclidean_recall": 0.864406779661017, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.71484375, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 734.9889526367188, | |
| "eval_Qnli-dev_manhattan_ap": 0.7578518420666434, | |
| "eval_Qnli-dev_manhattan_f1": 0.7160940325497287, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 797.8458251953125, | |
| "eval_Qnli-dev_manhattan_precision": 0.6246056782334385, | |
| "eval_Qnli-dev_manhattan_recall": 0.8389830508474576, | |
| "eval_Qnli-dev_max_accuracy": 0.71484375, | |
| "eval_Qnli-dev_max_accuracy_threshold": 734.9889526367188, | |
| "eval_Qnli-dev_max_ap": 0.7578518420666434, | |
| "eval_Qnli-dev_max_f1": 0.7160940325497287, | |
| "eval_Qnli-dev_max_f1_threshold": 797.8458251953125, | |
| "eval_Qnli-dev_max_precision": 0.6246056782334385, | |
| "eval_Qnli-dev_max_recall": 0.9110169491525424, | |
| "eval_allNLI-dev_cosine_accuracy": 0.712890625, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8226721286773682, | |
| "eval_allNLI-dev_cosine_ap": 0.608903927832523, | |
| "eval_allNLI-dev_cosine_f1": 0.6211764705882353, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6668639183044434, | |
| "eval_allNLI-dev_cosine_precision": 0.5238095238095238, | |
| "eval_allNLI-dev_cosine_recall": 0.7630057803468208, | |
| "eval_allNLI-dev_dot_accuracy": 0.72265625, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 701.8555908203125, | |
| "eval_allNLI-dev_dot_ap": 0.6006292150580212, | |
| "eval_allNLI-dev_dot_f1": 0.6206896551724138, | |
| "eval_allNLI-dev_dot_f1_threshold": 543.5947265625, | |
| "eval_allNLI-dev_dot_precision": 0.4948453608247423, | |
| "eval_allNLI-dev_dot_recall": 0.8323699421965318, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.716796875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 21.63890266418457, | |
| "eval_allNLI-dev_euclidean_ap": 0.6064044650997461, | |
| "eval_allNLI-dev_euclidean_f1": 0.6169354838709676, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 25.579940795898438, | |
| "eval_allNLI-dev_euclidean_precision": 0.47368421052631576, | |
| "eval_allNLI-dev_euclidean_recall": 0.884393063583815, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.71484375, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 673.708251953125, | |
| "eval_allNLI-dev_manhattan_ap": 0.6078798861215969, | |
| "eval_allNLI-dev_manhattan_f1": 0.6170212765957447, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 779.1580200195312, | |
| "eval_allNLI-dev_manhattan_precision": 0.4882154882154882, | |
| "eval_allNLI-dev_manhattan_recall": 0.838150289017341, | |
| "eval_allNLI-dev_max_accuracy": 0.72265625, | |
| "eval_allNLI-dev_max_accuracy_threshold": 701.8555908203125, | |
| "eval_allNLI-dev_max_ap": 0.608903927832523, | |
| "eval_allNLI-dev_max_f1": 0.6211764705882353, | |
| "eval_allNLI-dev_max_f1_threshold": 779.1580200195312, | |
| "eval_allNLI-dev_max_precision": 0.5238095238095238, | |
| "eval_allNLI-dev_max_recall": 0.884393063583815, | |
| "eval_sequential_score": 0.7578518420666434, | |
| "eval_sts-test_pearson_cosine": 0.9085589003585355, | |
| "eval_sts-test_pearson_dot": 0.8934066815875845, | |
| "eval_sts-test_pearson_euclidean": 0.9132129682245754, | |
| "eval_sts-test_pearson_manhattan": 0.9127682865746231, | |
| "eval_sts-test_pearson_max": 0.9132129682245754, | |
| "eval_sts-test_spearman_cosine": 0.9109138499261769, | |
| "eval_sts-test_spearman_dot": 0.8920668114399275, | |
| "eval_sts-test_spearman_euclidean": 0.912336089764457, | |
| "eval_sts-test_spearman_manhattan": 0.9119349842086059, | |
| "eval_sts-test_spearman_max": 0.912336089764457, | |
| "eval_vitaminc-pairs_loss": 2.359689474105835, | |
| "eval_vitaminc-pairs_runtime": 4.4921, | |
| "eval_vitaminc-pairs_samples_per_second": 28.494, | |
| "eval_vitaminc-pairs_steps_per_second": 0.223, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_negation-triplets_loss": 0.6828347444534302, | |
| "eval_negation-triplets_runtime": 3.0767, | |
| "eval_negation-triplets_samples_per_second": 41.603, | |
| "eval_negation-triplets_steps_per_second": 0.325, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_scitail-pairs-pos_loss": 0.020232411101460457, | |
| "eval_scitail-pairs-pos_runtime": 2.6103, | |
| "eval_scitail-pairs-pos_samples_per_second": 49.036, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.383, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_scitail-pairs-qa_loss": 0.0017561395652592182, | |
| "eval_scitail-pairs-qa_runtime": 2.2409, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.12, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.446, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_xsum-pairs_loss": 0.009539155289530754, | |
| "eval_xsum-pairs_runtime": 3.0343, | |
| "eval_xsum-pairs_samples_per_second": 42.184, | |
| "eval_xsum-pairs_steps_per_second": 0.33, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_sciq_pairs_loss": 0.07515428215265274, | |
| "eval_sciq_pairs_runtime": 3.8288, | |
| "eval_sciq_pairs_samples_per_second": 33.431, | |
| "eval_sciq_pairs_steps_per_second": 0.261, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_qasc_pairs_loss": 0.16715534031391144, | |
| "eval_qasc_pairs_runtime": 2.0736, | |
| "eval_qasc_pairs_samples_per_second": 61.729, | |
| "eval_qasc_pairs_steps_per_second": 0.482, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_openbookqa_pairs_loss": 0.5365710854530334, | |
| "eval_openbookqa_pairs_runtime": 2.2749, | |
| "eval_openbookqa_pairs_samples_per_second": 56.267, | |
| "eval_openbookqa_pairs_steps_per_second": 0.44, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_msmarco_pairs_loss": 0.183290034532547, | |
| "eval_msmarco_pairs_runtime": 2.2376, | |
| "eval_msmarco_pairs_samples_per_second": 57.204, | |
| "eval_msmarco_pairs_steps_per_second": 0.447, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_nq_pairs_loss": 0.13633984327316284, | |
| "eval_nq_pairs_runtime": 2.7168, | |
| "eval_nq_pairs_samples_per_second": 47.115, | |
| "eval_nq_pairs_steps_per_second": 0.368, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_trivia_pairs_loss": 0.13907591998577118, | |
| "eval_trivia_pairs_runtime": 3.7638, | |
| "eval_trivia_pairs_samples_per_second": 34.008, | |
| "eval_trivia_pairs_steps_per_second": 0.266, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_gooaq_pairs_loss": 0.15382522344589233, | |
| "eval_gooaq_pairs_runtime": 2.1349, | |
| "eval_gooaq_pairs_samples_per_second": 59.955, | |
| "eval_gooaq_pairs_steps_per_second": 0.468, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_paws-pos_loss": 0.02764580212533474, | |
| "eval_paws-pos_runtime": 2.299, | |
| "eval_paws-pos_samples_per_second": 55.676, | |
| "eval_paws-pos_steps_per_second": 0.435, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25116519937856036, | |
| "eval_global_dataset_loss": 0.26656147837638855, | |
| "eval_global_dataset_runtime": 10.0817, | |
| "eval_global_dataset_samples_per_second": 41.263, | |
| "eval_global_dataset_steps_per_second": 0.397, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.2589331952356292, | |
| "grad_norm": 8.639649391174316, | |
| "learning_rate": 1.1529411764705883e-05, | |
| "loss": 0.1781, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.26929052304505435, | |
| "grad_norm": 14.192313194274902, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.2177, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.2796478508544795, | |
| "grad_norm": 0.47864726185798645, | |
| "learning_rate": 1.2470588235294119e-05, | |
| "loss": 0.5771, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.29000517866390474, | |
| "grad_norm": 94.00303649902344, | |
| "learning_rate": 1.291764705882353e-05, | |
| "loss": 2.2303, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3003625064733299, | |
| "grad_norm": 284.8737487792969, | |
| "learning_rate": 1.3364705882352942e-05, | |
| "loss": 1.0045, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_Qnli-dev_cosine_accuracy": 0.6953125, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.8107659816741943, | |
| "eval_Qnli-dev_cosine_ap": 0.7306763373744987, | |
| "eval_Qnli-dev_cosine_f1": 0.6955074875207986, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.7081253528594971, | |
| "eval_Qnli-dev_cosine_precision": 0.5726027397260274, | |
| "eval_Qnli-dev_cosine_recall": 0.885593220338983, | |
| "eval_Qnli-dev_dot_accuracy": 0.671875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 875.9421997070312, | |
| "eval_Qnli-dev_dot_ap": 0.6876970673529026, | |
| "eval_Qnli-dev_dot_f1": 0.6821192052980133, | |
| "eval_Qnli-dev_dot_f1_threshold": 786.2505493164062, | |
| "eval_Qnli-dev_dot_precision": 0.5597826086956522, | |
| "eval_Qnli-dev_dot_recall": 0.8728813559322034, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.69921875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 21.132396697998047, | |
| "eval_Qnli-dev_euclidean_ap": 0.7325235937497143, | |
| "eval_Qnli-dev_euclidean_f1": 0.6955074875207986, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 25.694360733032227, | |
| "eval_Qnli-dev_euclidean_precision": 0.5726027397260274, | |
| "eval_Qnli-dev_euclidean_recall": 0.885593220338983, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.701171875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 650.69775390625, | |
| "eval_Qnli-dev_manhattan_ap": 0.73351057649253, | |
| "eval_Qnli-dev_manhattan_f1": 0.6923076923076924, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 790.9528198242188, | |
| "eval_Qnli-dev_manhattan_precision": 0.5718232044198895, | |
| "eval_Qnli-dev_manhattan_recall": 0.8771186440677966, | |
| "eval_Qnli-dev_max_accuracy": 0.701171875, | |
| "eval_Qnli-dev_max_accuracy_threshold": 875.9421997070312, | |
| "eval_Qnli-dev_max_ap": 0.73351057649253, | |
| "eval_Qnli-dev_max_f1": 0.6955074875207986, | |
| "eval_Qnli-dev_max_f1_threshold": 790.9528198242188, | |
| "eval_Qnli-dev_max_precision": 0.5726027397260274, | |
| "eval_Qnli-dev_max_recall": 0.885593220338983, | |
| "eval_allNLI-dev_cosine_accuracy": 0.71875, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8345531225204468, | |
| "eval_allNLI-dev_cosine_ap": 0.5833256810054208, | |
| "eval_allNLI-dev_cosine_f1": 0.6093366093366094, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7519584894180298, | |
| "eval_allNLI-dev_cosine_precision": 0.5299145299145299, | |
| "eval_allNLI-dev_cosine_recall": 0.7167630057803468, | |
| "eval_allNLI-dev_dot_accuracy": 0.71875, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 932.83544921875, | |
| "eval_allNLI-dev_dot_ap": 0.5730668161963208, | |
| "eval_allNLI-dev_dot_f1": 0.6140350877192983, | |
| "eval_allNLI-dev_dot_f1_threshold": 790.3121337890625, | |
| "eval_allNLI-dev_dot_precision": 0.49469964664310956, | |
| "eval_allNLI-dev_dot_recall": 0.8092485549132948, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.712890625, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 16.462337493896484, | |
| "eval_allNLI-dev_euclidean_ap": 0.5830290393354319, | |
| "eval_allNLI-dev_euclidean_f1": 0.6080760095011876, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 23.817108154296875, | |
| "eval_allNLI-dev_euclidean_precision": 0.5161290322580645, | |
| "eval_allNLI-dev_euclidean_recall": 0.7398843930635838, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.71484375, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 514.4776611328125, | |
| "eval_allNLI-dev_manhattan_ap": 0.5824345257218883, | |
| "eval_allNLI-dev_manhattan_f1": 0.6029411764705882, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 725.9110717773438, | |
| "eval_allNLI-dev_manhattan_precision": 0.5234042553191489, | |
| "eval_allNLI-dev_manhattan_recall": 0.7109826589595376, | |
| "eval_allNLI-dev_max_accuracy": 0.71875, | |
| "eval_allNLI-dev_max_accuracy_threshold": 932.83544921875, | |
| "eval_allNLI-dev_max_ap": 0.5833256810054208, | |
| "eval_allNLI-dev_max_f1": 0.6140350877192983, | |
| "eval_allNLI-dev_max_f1_threshold": 790.3121337890625, | |
| "eval_allNLI-dev_max_precision": 0.5299145299145299, | |
| "eval_allNLI-dev_max_recall": 0.8092485549132948, | |
| "eval_sequential_score": 0.73351057649253, | |
| "eval_sts-test_pearson_cosine": 0.911958388742002, | |
| "eval_sts-test_pearson_dot": 0.8881053452310657, | |
| "eval_sts-test_pearson_euclidean": 0.9250703199093523, | |
| "eval_sts-test_pearson_manhattan": 0.9254282934479543, | |
| "eval_sts-test_pearson_max": 0.9254282934479543, | |
| "eval_sts-test_spearman_cosine": 0.9182240579769849, | |
| "eval_sts-test_spearman_dot": 0.8777027753148232, | |
| "eval_sts-test_spearman_euclidean": 0.9183138737585973, | |
| "eval_sts-test_spearman_manhattan": 0.9189913183535404, | |
| "eval_sts-test_spearman_max": 0.9189913183535404, | |
| "eval_vitaminc-pairs_loss": 3.1416079998016357, | |
| "eval_vitaminc-pairs_runtime": 4.5, | |
| "eval_vitaminc-pairs_samples_per_second": 28.445, | |
| "eval_vitaminc-pairs_steps_per_second": 0.222, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_negation-triplets_loss": 0.7199142575263977, | |
| "eval_negation-triplets_runtime": 3.1016, | |
| "eval_negation-triplets_samples_per_second": 41.27, | |
| "eval_negation-triplets_steps_per_second": 0.322, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_scitail-pairs-pos_loss": 0.020503610372543335, | |
| "eval_scitail-pairs-pos_runtime": 2.6474, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.35, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.378, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_scitail-pairs-qa_loss": 0.014284193515777588, | |
| "eval_scitail-pairs-qa_runtime": 2.2281, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.447, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.449, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_xsum-pairs_loss": 0.020332960411906242, | |
| "eval_xsum-pairs_runtime": 3.0646, | |
| "eval_xsum-pairs_samples_per_second": 41.767, | |
| "eval_xsum-pairs_steps_per_second": 0.326, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_sciq_pairs_loss": 0.07365372776985168, | |
| "eval_sciq_pairs_runtime": 3.9037, | |
| "eval_sciq_pairs_samples_per_second": 32.789, | |
| "eval_sciq_pairs_steps_per_second": 0.256, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_qasc_pairs_loss": 0.9374014735221863, | |
| "eval_qasc_pairs_runtime": 2.0898, | |
| "eval_qasc_pairs_samples_per_second": 61.249, | |
| "eval_qasc_pairs_steps_per_second": 0.479, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_openbookqa_pairs_loss": 0.5403007864952087, | |
| "eval_openbookqa_pairs_runtime": 2.1959, | |
| "eval_openbookqa_pairs_samples_per_second": 58.291, | |
| "eval_openbookqa_pairs_steps_per_second": 0.455, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_msmarco_pairs_loss": 2.8998327255249023, | |
| "eval_msmarco_pairs_runtime": 2.2455, | |
| "eval_msmarco_pairs_samples_per_second": 57.004, | |
| "eval_msmarco_pairs_steps_per_second": 0.445, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_nq_pairs_loss": 0.1996317207813263, | |
| "eval_nq_pairs_runtime": 2.7271, | |
| "eval_nq_pairs_samples_per_second": 46.936, | |
| "eval_nq_pairs_steps_per_second": 0.367, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_trivia_pairs_loss": 0.16345469653606415, | |
| "eval_trivia_pairs_runtime": 3.7445, | |
| "eval_trivia_pairs_samples_per_second": 34.183, | |
| "eval_trivia_pairs_steps_per_second": 0.267, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_gooaq_pairs_loss": 1.6607106924057007, | |
| "eval_gooaq_pairs_runtime": 2.1231, | |
| "eval_gooaq_pairs_samples_per_second": 60.289, | |
| "eval_gooaq_pairs_steps_per_second": 0.471, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_paws-pos_loss": 0.026082171127200127, | |
| "eval_paws-pos_runtime": 2.2328, | |
| "eval_paws-pos_samples_per_second": 57.327, | |
| "eval_paws-pos_steps_per_second": 0.448, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3013982392542724, | |
| "eval_global_dataset_loss": 0.4393865168094635, | |
| "eval_global_dataset_runtime": 10.0711, | |
| "eval_global_dataset_samples_per_second": 41.307, | |
| "eval_global_dataset_steps_per_second": 0.397, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.31071983428275507, | |
| "grad_norm": 2.5726535320281982, | |
| "learning_rate": 1.3835294117647059e-05, | |
| "loss": 0.5632, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.32107716209218023, | |
| "grad_norm": 65.14546966552734, | |
| "learning_rate": 1.4305882352941177e-05, | |
| "loss": 0.2533, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.3314344899016054, | |
| "grad_norm": 4.207058906555176, | |
| "learning_rate": 1.4776470588235294e-05, | |
| "loss": 0.2559, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.34179181771103057, | |
| "grad_norm": 278.2864990234375, | |
| "learning_rate": 1.5223529411764707e-05, | |
| "loss": 0.2664, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_Qnli-dev_cosine_accuracy": 0.69140625, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.677927553653717, | |
| "eval_Qnli-dev_cosine_ap": 0.7135273513673487, | |
| "eval_Qnli-dev_cosine_f1": 0.7079037800687284, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6066854000091553, | |
| "eval_Qnli-dev_cosine_precision": 0.5953757225433526, | |
| "eval_Qnli-dev_cosine_recall": 0.8728813559322034, | |
| "eval_Qnli-dev_dot_accuracy": 0.669921875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 615.3697509765625, | |
| "eval_Qnli-dev_dot_ap": 0.6697900654196967, | |
| "eval_Qnli-dev_dot_f1": 0.6946308724832215, | |
| "eval_Qnli-dev_dot_f1_threshold": 583.5701293945312, | |
| "eval_Qnli-dev_dot_precision": 0.575, | |
| "eval_Qnli-dev_dot_recall": 0.8771186440677966, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.70703125, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 24.364826202392578, | |
| "eval_Qnli-dev_euclidean_ap": 0.720573540901372, | |
| "eval_Qnli-dev_euclidean_f1": 0.7003367003367004, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 28.203102111816406, | |
| "eval_Qnli-dev_euclidean_precision": 0.5810055865921788, | |
| "eval_Qnli-dev_euclidean_recall": 0.8813559322033898, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.705078125, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 754.4404296875, | |
| "eval_Qnli-dev_manhattan_ap": 0.7236352677013607, | |
| "eval_Qnli-dev_manhattan_f1": 0.701168614357262, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 882.8988037109375, | |
| "eval_Qnli-dev_manhattan_precision": 0.5785123966942148, | |
| "eval_Qnli-dev_manhattan_recall": 0.8898305084745762, | |
| "eval_Qnli-dev_max_accuracy": 0.70703125, | |
| "eval_Qnli-dev_max_accuracy_threshold": 754.4404296875, | |
| "eval_Qnli-dev_max_ap": 0.7236352677013607, | |
| "eval_Qnli-dev_max_f1": 0.7079037800687284, | |
| "eval_Qnli-dev_max_f1_threshold": 882.8988037109375, | |
| "eval_Qnli-dev_max_precision": 0.5953757225433526, | |
| "eval_Qnli-dev_max_recall": 0.8898305084745762, | |
| "eval_allNLI-dev_cosine_accuracy": 0.7109375, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8685251474380493, | |
| "eval_allNLI-dev_cosine_ap": 0.5715847909509861, | |
| "eval_allNLI-dev_cosine_f1": 0.5822222222222222, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6784489154815674, | |
| "eval_allNLI-dev_cosine_precision": 0.4729241877256318, | |
| "eval_allNLI-dev_cosine_recall": 0.7572254335260116, | |
| "eval_allNLI-dev_dot_accuracy": 0.705078125, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 886.351318359375, | |
| "eval_allNLI-dev_dot_ap": 0.5644148142455192, | |
| "eval_allNLI-dev_dot_f1": 0.5940170940170941, | |
| "eval_allNLI-dev_dot_f1_threshold": 659.7857666015625, | |
| "eval_allNLI-dev_dot_precision": 0.4711864406779661, | |
| "eval_allNLI-dev_dot_recall": 0.8034682080924855, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.708984375, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 15.756305694580078, | |
| "eval_allNLI-dev_euclidean_ap": 0.5670650775218166, | |
| "eval_allNLI-dev_euclidean_f1": 0.5840000000000001, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 26.798587799072266, | |
| "eval_allNLI-dev_euclidean_precision": 0.44648318042813456, | |
| "eval_allNLI-dev_euclidean_recall": 0.8439306358381503, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.712890625, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 551.571533203125, | |
| "eval_allNLI-dev_manhattan_ap": 0.5708799354607844, | |
| "eval_allNLI-dev_manhattan_f1": 0.5863453815261045, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 831.1856689453125, | |
| "eval_allNLI-dev_manhattan_precision": 0.4492307692307692, | |
| "eval_allNLI-dev_manhattan_recall": 0.8439306358381503, | |
| "eval_allNLI-dev_max_accuracy": 0.712890625, | |
| "eval_allNLI-dev_max_accuracy_threshold": 886.351318359375, | |
| "eval_allNLI-dev_max_ap": 0.5715847909509861, | |
| "eval_allNLI-dev_max_f1": 0.5940170940170941, | |
| "eval_allNLI-dev_max_f1_threshold": 831.1856689453125, | |
| "eval_allNLI-dev_max_precision": 0.4729241877256318, | |
| "eval_allNLI-dev_max_recall": 0.8439306358381503, | |
| "eval_sequential_score": 0.7236352677013607, | |
| "eval_sts-test_pearson_cosine": 0.9021470099112117, | |
| "eval_sts-test_pearson_dot": 0.8917251782363848, | |
| "eval_sts-test_pearson_euclidean": 0.9135854372994908, | |
| "eval_sts-test_pearson_manhattan": 0.9142235212520329, | |
| "eval_sts-test_pearson_max": 0.9142235212520329, | |
| "eval_sts-test_spearman_cosine": 0.907700664607013, | |
| "eval_sts-test_spearman_dot": 0.8887621545424798, | |
| "eval_sts-test_spearman_euclidean": 0.9072292995707775, | |
| "eval_sts-test_spearman_manhattan": 0.9078543851898925, | |
| "eval_sts-test_spearman_max": 0.9078543851898925, | |
| "eval_vitaminc-pairs_loss": 2.5545308589935303, | |
| "eval_vitaminc-pairs_runtime": 4.4484, | |
| "eval_vitaminc-pairs_samples_per_second": 28.775, | |
| "eval_vitaminc-pairs_steps_per_second": 0.225, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_negation-triplets_loss": 0.792955756187439, | |
| "eval_negation-triplets_runtime": 3.0828, | |
| "eval_negation-triplets_samples_per_second": 41.521, | |
| "eval_negation-triplets_steps_per_second": 0.324, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_scitail-pairs-pos_loss": 0.011221353895962238, | |
| "eval_scitail-pairs-pos_runtime": 2.6324, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.624, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.38, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_scitail-pairs-qa_loss": 0.008626868017017841, | |
| "eval_scitail-pairs-qa_runtime": 2.1929, | |
| "eval_scitail-pairs-qa_samples_per_second": 58.369, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.456, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_xsum-pairs_loss": 0.004584914073348045, | |
| "eval_xsum-pairs_runtime": 3.0494, | |
| "eval_xsum-pairs_samples_per_second": 41.975, | |
| "eval_xsum-pairs_steps_per_second": 0.328, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_sciq_pairs_loss": 0.07755079865455627, | |
| "eval_sciq_pairs_runtime": 3.8446, | |
| "eval_sciq_pairs_samples_per_second": 33.294, | |
| "eval_sciq_pairs_steps_per_second": 0.26, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_qasc_pairs_loss": 0.2504812777042389, | |
| "eval_qasc_pairs_runtime": 2.0696, | |
| "eval_qasc_pairs_samples_per_second": 61.849, | |
| "eval_qasc_pairs_steps_per_second": 0.483, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_openbookqa_pairs_loss": 0.615034818649292, | |
| "eval_openbookqa_pairs_runtime": 2.1876, | |
| "eval_openbookqa_pairs_samples_per_second": 58.512, | |
| "eval_openbookqa_pairs_steps_per_second": 0.457, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_msmarco_pairs_loss": 0.12970499694347382, | |
| "eval_msmarco_pairs_runtime": 2.2547, | |
| "eval_msmarco_pairs_samples_per_second": 56.771, | |
| "eval_msmarco_pairs_steps_per_second": 0.444, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_nq_pairs_loss": 0.11113697290420532, | |
| "eval_nq_pairs_runtime": 2.7434, | |
| "eval_nq_pairs_samples_per_second": 46.658, | |
| "eval_nq_pairs_steps_per_second": 0.365, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_trivia_pairs_loss": 0.13746751844882965, | |
| "eval_trivia_pairs_runtime": 3.7502, | |
| "eval_trivia_pairs_samples_per_second": 34.131, | |
| "eval_trivia_pairs_steps_per_second": 0.267, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_gooaq_pairs_loss": 0.15882055461406708, | |
| "eval_gooaq_pairs_runtime": 2.1544, | |
| "eval_gooaq_pairs_samples_per_second": 59.413, | |
| "eval_gooaq_pairs_steps_per_second": 0.464, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_paws-pos_loss": 0.024272503331303596, | |
| "eval_paws-pos_runtime": 2.2216, | |
| "eval_paws-pos_samples_per_second": 57.616, | |
| "eval_paws-pos_steps_per_second": 0.45, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.3516312791299845, | |
| "eval_global_dataset_loss": 0.19072183966636658, | |
| "eval_global_dataset_runtime": 10.0803, | |
| "eval_global_dataset_samples_per_second": 41.269, | |
| "eval_global_dataset_steps_per_second": 0.397, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.35214914552045573, | |
| "grad_norm": 13.81802749633789, | |
| "learning_rate": 1.5694117647058825e-05, | |
| "loss": 0.2108, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3625064733298809, | |
| "grad_norm": 1107.8421630859375, | |
| "learning_rate": 1.6164705882352942e-05, | |
| "loss": 0.2936, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.37286380113930606, | |
| "grad_norm": 9.161267280578613, | |
| "learning_rate": 1.6635294117647062e-05, | |
| "loss": 1.13, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.3832211289487312, | |
| "grad_norm": 523.1478881835938, | |
| "learning_rate": 1.7105882352941176e-05, | |
| "loss": 0.2598, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.3935784567581564, | |
| "grad_norm": 0.9771941304206848, | |
| "learning_rate": 1.7576470588235296e-05, | |
| "loss": 0.1599, | |
| "step": 760 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 3862, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 387, | |
| "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": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |