diff --git "a/checkpoint-258/trainer_state.json" "b/checkpoint-258/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-258/trainer_state.json"
@@ -0,0 +1,4155 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.20015515903801395,
+  "eval_steps": 20,
+  "global_step": 258,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0007757951900698216,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 6.6798,
+      "step": 1
+    },
+    {
+      "epoch": 0.0015515903801396431,
+      "grad_norm": 19.460369110107422,
+      "learning_rate": 4.112808460634547e-08,
+      "loss": 6.1418,
+      "step": 2
+    },
+    {
+      "epoch": 0.0023273855702094647,
+      "grad_norm": 38.97050857543945,
+      "learning_rate": 8.225616921269094e-08,
+      "loss": 6.8984,
+      "step": 3
+    },
+    {
+      "epoch": 0.0031031807602792862,
+      "grad_norm": 25.763484954833984,
+      "learning_rate": 1.2338425381903642e-07,
+      "loss": 7.1881,
+      "step": 4
+    },
+    {
+      "epoch": 0.003878975950349108,
+      "grad_norm": 11.17786979675293,
+      "learning_rate": 1.6451233842538187e-07,
+      "loss": 5.5114,
+      "step": 5
+    },
+    {
+      "epoch": 0.004654771140418929,
+      "grad_norm": 15.640860557556152,
+      "learning_rate": 2.0564042303172737e-07,
+      "loss": 6.2131,
+      "step": 6
+    },
+    {
+      "epoch": 0.005430566330488751,
+      "grad_norm": 26.026487350463867,
+      "learning_rate": 2.4676850763807285e-07,
+      "loss": 7.43,
+      "step": 7
+    },
+    {
+      "epoch": 0.0062063615205585725,
+      "grad_norm": Infinity,
+      "learning_rate": 2.4676850763807285e-07,
+      "loss": 9.8237,
+      "step": 8
+    },
+    {
+      "epoch": 0.0069821567106283944,
+      "grad_norm": 12.019311904907227,
+      "learning_rate": 2.878965922444183e-07,
+      "loss": 5.4554,
+      "step": 9
+    },
+    {
+      "epoch": 0.007757951900698216,
+      "grad_norm": 19.58979606628418,
+      "learning_rate": 3.2902467685076374e-07,
+      "loss": 6.6825,
+      "step": 10
+    },
+    {
+      "epoch": 0.008533747090768037,
+      "grad_norm": 11.048348426818848,
+      "learning_rate": 3.701527614571093e-07,
+      "loss": 5.6752,
+      "step": 11
+    },
+    {
+      "epoch": 0.009309542280837859,
+      "grad_norm": 10.552579879760742,
+      "learning_rate": 4.1128084606345474e-07,
+      "loss": 4.8126,
+      "step": 12
+    },
+    {
+      "epoch": 0.01008533747090768,
+      "grad_norm": 10.90353775024414,
+      "learning_rate": 4.524089306698002e-07,
+      "loss": 4.7993,
+      "step": 13
+    },
+    {
+      "epoch": 0.010861132660977503,
+      "grad_norm": 14.389288902282715,
+      "learning_rate": 4.935370152761457e-07,
+      "loss": 6.2433,
+      "step": 14
+    },
+    {
+      "epoch": 0.011636927851047323,
+      "grad_norm": 32.469276428222656,
+      "learning_rate": 5.346650998824911e-07,
+      "loss": 7.8983,
+      "step": 15
+    },
+    {
+      "epoch": 0.012412723041117145,
+      "grad_norm": 17.988374710083008,
+      "learning_rate": 5.757931844888366e-07,
+      "loss": 5.9294,
+      "step": 16
+    },
+    {
+      "epoch": 0.013188518231186967,
+      "grad_norm": 53.49104309082031,
+      "learning_rate": 6.169212690951821e-07,
+      "loss": 10.0069,
+      "step": 17
+    },
+    {
+      "epoch": 0.013964313421256789,
+      "grad_norm": 51.57087707519531,
+      "learning_rate": 6.580493537015275e-07,
+      "loss": 9.761,
+      "step": 18
+    },
+    {
+      "epoch": 0.01474010861132661,
+      "grad_norm": 10.312777519226074,
+      "learning_rate": 6.99177438307873e-07,
+      "loss": 4.7492,
+      "step": 19
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "grad_norm": 24.298805236816406,
+      "learning_rate": 7.403055229142186e-07,
+      "loss": 7.0924,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_Qnli-dev_cosine_accuracy": 0.58984375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9331285953521729,
+      "eval_Qnli-dev_cosine_ap": 0.5485925209297758,
+      "eval_Qnli-dev_cosine_f1": 0.6291834002677376,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.6565657258033752,
+      "eval_Qnli-dev_cosine_precision": 0.4598825831702544,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.541015625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 1048.2359619140625,
+      "eval_Qnli-dev_dot_ap": 0.4748490344531905,
+      "eval_Qnli-dev_dot_f1": 0.6300268096514745,
+      "eval_Qnli-dev_dot_f1_threshold": 380.322998046875,
+      "eval_Qnli-dev_dot_precision": 0.46078431372549017,
+      "eval_Qnli-dev_dot_recall": 0.9957627118644068,
+      "eval_Qnli-dev_euclidean_accuracy": 0.587890625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 9.073010444641113,
+      "eval_Qnli-dev_euclidean_ap": 0.5650640204703478,
+      "eval_Qnli-dev_euclidean_f1": 0.6332882273342355,
+      "eval_Qnli-dev_euclidean_f1_threshold": 18.225353240966797,
+      "eval_Qnli-dev_euclidean_precision": 0.46520874751491054,
+      "eval_Qnli-dev_euclidean_recall": 0.9915254237288136,
+      "eval_Qnli-dev_manhattan_accuracy": 0.6171875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 155.41647338867188,
+      "eval_Qnli-dev_manhattan_ap": 0.6018518855824669,
+      "eval_Qnli-dev_manhattan_f1": 0.6291834002677376,
+      "eval_Qnli-dev_manhattan_f1_threshold": 286.5093994140625,
+      "eval_Qnli-dev_manhattan_precision": 0.4598825831702544,
+      "eval_Qnli-dev_manhattan_recall": 0.9957627118644068,
+      "eval_Qnli-dev_max_accuracy": 0.6171875,
+      "eval_Qnli-dev_max_accuracy_threshold": 1048.2359619140625,
+      "eval_Qnli-dev_max_ap": 0.6018518855824669,
+      "eval_Qnli-dev_max_f1": 0.6332882273342355,
+      "eval_Qnli-dev_max_f1_threshold": 380.322998046875,
+      "eval_Qnli-dev_max_precision": 0.46520874751491054,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.666015625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9902692437171936,
+      "eval_allNLI-dev_cosine_ap": 0.36630271296437167,
+      "eval_allNLI-dev_cosine_f1": 0.5088235294117648,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.7946319580078125,
+      "eval_allNLI-dev_cosine_precision": 0.34122287968441817,
+      "eval_allNLI-dev_cosine_recall": 1.0,
+      "eval_allNLI-dev_dot_accuracy": 0.662109375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 940.7398681640625,
+      "eval_allNLI-dev_dot_ap": 0.33409368825504626,
+      "eval_allNLI-dev_dot_f1": 0.5081240768094535,
+      "eval_allNLI-dev_dot_f1_threshold": 585.4312744140625,
+      "eval_allNLI-dev_dot_precision": 0.3412698412698413,
+      "eval_allNLI-dev_dot_recall": 0.9942196531791907,
+      "eval_allNLI-dev_euclidean_accuracy": 0.666015625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.293120384216309,
+      "eval_allNLI-dev_euclidean_ap": 0.36165210170894113,
+      "eval_allNLI-dev_euclidean_f1": 0.5111441307578009,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.931983947753906,
+      "eval_allNLI-dev_euclidean_precision": 0.344,
+      "eval_allNLI-dev_euclidean_recall": 0.9942196531791907,
+      "eval_allNLI-dev_manhattan_accuracy": 0.666015625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 74.05484008789062,
+      "eval_allNLI-dev_manhattan_ap": 0.37228608061825896,
+      "eval_allNLI-dev_manhattan_f1": 0.5081240768094535,
+      "eval_allNLI-dev_manhattan_f1_threshold": 240.91050720214844,
+      "eval_allNLI-dev_manhattan_precision": 0.3412698412698413,
+      "eval_allNLI-dev_manhattan_recall": 0.9942196531791907,
+      "eval_allNLI-dev_max_accuracy": 0.666015625,
+      "eval_allNLI-dev_max_accuracy_threshold": 940.7398681640625,
+      "eval_allNLI-dev_max_ap": 0.37228608061825896,
+      "eval_allNLI-dev_max_f1": 0.5111441307578009,
+      "eval_allNLI-dev_max_f1_threshold": 585.4312744140625,
+      "eval_allNLI-dev_max_precision": 0.344,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6018518855824669,
+      "eval_sts-test_pearson_cosine": 0.14468011325378688,
+      "eval_sts-test_pearson_dot": 0.12373166012012136,
+      "eval_sts-test_pearson_euclidean": 0.15040853567589774,
+      "eval_sts-test_pearson_manhattan": 0.14680735741048356,
+      "eval_sts-test_pearson_max": 0.15040853567589774,
+      "eval_sts-test_spearman_cosine": 0.1980088477205014,
+      "eval_sts-test_spearman_dot": 0.12132071438334546,
+      "eval_sts-test_spearman_euclidean": 0.17802306863688658,
+      "eval_sts-test_spearman_manhattan": 0.17413140944376768,
+      "eval_sts-test_spearman_max": 0.1980088477205014,
+      "eval_vitaminc-pairs_loss": 4.239284515380859,
+      "eval_vitaminc-pairs_runtime": 5.4781,
+      "eval_vitaminc-pairs_samples_per_second": 23.366,
+      "eval_vitaminc-pairs_steps_per_second": 0.365,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_negation-triplets_loss": 5.499993801116943,
+      "eval_negation-triplets_runtime": 1.0772,
+      "eval_negation-triplets_samples_per_second": 118.828,
+      "eval_negation-triplets_steps_per_second": 1.857,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_scitail-pairs-pos_loss": 2.806995391845703,
+      "eval_scitail-pairs-pos_runtime": 1.221,
+      "eval_scitail-pairs-pos_samples_per_second": 104.828,
+      "eval_scitail-pairs-pos_steps_per_second": 1.638,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_scitail-pairs-qa_loss": 4.214225769042969,
+      "eval_scitail-pairs-qa_runtime": 0.7726,
+      "eval_scitail-pairs-qa_samples_per_second": 165.67,
+      "eval_scitail-pairs-qa_steps_per_second": 2.589,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_xsum-pairs_loss": 5.982180118560791,
+      "eval_xsum-pairs_runtime": 3.9493,
+      "eval_xsum-pairs_samples_per_second": 32.411,
+      "eval_xsum-pairs_steps_per_second": 0.506,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_sciq_pairs_loss": 0.733872652053833,
+      "eval_sciq_pairs_runtime": 5.8101,
+      "eval_sciq_pairs_samples_per_second": 22.031,
+      "eval_sciq_pairs_steps_per_second": 0.344,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_qasc_pairs_loss": 8.005105018615723,
+      "eval_qasc_pairs_runtime": 0.89,
+      "eval_qasc_pairs_samples_per_second": 143.822,
+      "eval_qasc_pairs_steps_per_second": 2.247,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_openbookqa_pairs_loss": 5.631566047668457,
+      "eval_openbookqa_pairs_runtime": 0.861,
+      "eval_openbookqa_pairs_samples_per_second": 148.656,
+      "eval_openbookqa_pairs_steps_per_second": 2.323,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_msmarco_pairs_loss": 10.92141342163086,
+      "eval_msmarco_pairs_runtime": 1.9495,
+      "eval_msmarco_pairs_samples_per_second": 65.659,
+      "eval_msmarco_pairs_steps_per_second": 1.026,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_nq_pairs_loss": 9.838966369628906,
+      "eval_nq_pairs_runtime": 3.3145,
+      "eval_nq_pairs_samples_per_second": 38.618,
+      "eval_nq_pairs_steps_per_second": 0.603,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_trivia_pairs_loss": 7.321401596069336,
+      "eval_trivia_pairs_runtime": 4.4673,
+      "eval_trivia_pairs_samples_per_second": 28.653,
+      "eval_trivia_pairs_steps_per_second": 0.448,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_gooaq_pairs_loss": 7.844480037689209,
+      "eval_gooaq_pairs_runtime": 1.4628,
+      "eval_gooaq_pairs_samples_per_second": 87.505,
+      "eval_gooaq_pairs_steps_per_second": 1.367,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_paws-pos_loss": 2.474581718444824,
+      "eval_paws-pos_runtime": 1.0227,
+      "eval_paws-pos_samples_per_second": 125.159,
+      "eval_paws-pos_steps_per_second": 1.956,
+      "step": 20
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "eval_global_dataset_loss": 5.820813179016113,
+      "eval_global_dataset_runtime": 19.3774,
+      "eval_global_dataset_samples_per_second": 21.468,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 20
+    },
+    {
+      "epoch": 0.016291698991466253,
+      "grad_norm": 16.32306671142578,
+      "learning_rate": 7.81433607520564e-07,
+      "loss": 6.3011,
+      "step": 21
+    },
+    {
+      "epoch": 0.017067494181536073,
+      "grad_norm": 10.361891746520996,
+      "learning_rate": 8.225616921269095e-07,
+      "loss": 5.6518,
+      "step": 22
+    },
+    {
+      "epoch": 0.017843289371605897,
+      "grad_norm": 18.645973205566406,
+      "learning_rate": 8.636897767332549e-07,
+      "loss": 6.5445,
+      "step": 23
+    },
+    {
+      "epoch": 0.018619084561675717,
+      "grad_norm": 14.296663284301758,
+      "learning_rate": 9.048178613396004e-07,
+      "loss": 6.0762,
+      "step": 24
+    },
+    {
+      "epoch": 0.019394879751745538,
+      "grad_norm": 12.288186073303223,
+      "learning_rate": 9.459459459459459e-07,
+      "loss": 4.9043,
+      "step": 25
+    },
+    {
+      "epoch": 0.02017067494181536,
+      "grad_norm": 48.58110046386719,
+      "learning_rate": 9.870740305522914e-07,
+      "loss": 10.3442,
+      "step": 26
+    },
+    {
+      "epoch": 0.02094647013188518,
+      "grad_norm": 17.945547103881836,
+      "learning_rate": 1.0282021151586369e-06,
+      "loss": 6.3305,
+      "step": 27
+    },
+    {
+      "epoch": 0.021722265321955005,
+      "grad_norm": 16.731460571289062,
+      "learning_rate": 1.0693301997649822e-06,
+      "loss": 6.3366,
+      "step": 28
+    },
+    {
+      "epoch": 0.022498060512024826,
+      "grad_norm": 10.825992584228516,
+      "learning_rate": 1.1104582843713277e-06,
+      "loss": 5.6012,
+      "step": 29
+    },
+    {
+      "epoch": 0.023273855702094646,
+      "grad_norm": 13.152975082397461,
+      "learning_rate": 1.1515863689776732e-06,
+      "loss": 5.9001,
+      "step": 30
+    },
+    {
+      "epoch": 0.02404965089216447,
+      "grad_norm": 17.892440795898438,
+      "learning_rate": 1.1927144535840187e-06,
+      "loss": 6.3815,
+      "step": 31
+    },
+    {
+      "epoch": 0.02482544608223429,
+      "grad_norm": 10.021342277526855,
+      "learning_rate": 1.2338425381903642e-06,
+      "loss": 4.8941,
+      "step": 32
+    },
+    {
+      "epoch": 0.02560124127230411,
+      "grad_norm": 17.672340393066406,
+      "learning_rate": 1.2749706227967097e-06,
+      "loss": 6.2971,
+      "step": 33
+    },
+    {
+      "epoch": 0.026377036462373934,
+      "grad_norm": 13.532915115356445,
+      "learning_rate": 1.316098707403055e-06,
+      "loss": 5.6313,
+      "step": 34
+    },
+    {
+      "epoch": 0.027152831652443754,
+      "grad_norm": 13.363771438598633,
+      "learning_rate": 1.3572267920094007e-06,
+      "loss": 5.6635,
+      "step": 35
+    },
+    {
+      "epoch": 0.027928626842513578,
+      "grad_norm": 11.951986312866211,
+      "learning_rate": 1.398354876615746e-06,
+      "loss": 6.0632,
+      "step": 36
+    },
+    {
+      "epoch": 0.028704422032583398,
+      "grad_norm": 12.715692520141602,
+      "learning_rate": 1.4394829612220915e-06,
+      "loss": 6.0125,
+      "step": 37
+    },
+    {
+      "epoch": 0.02948021722265322,
+      "grad_norm": 44.77717208862305,
+      "learning_rate": 1.4806110458284372e-06,
+      "loss": 7.9108,
+      "step": 38
+    },
+    {
+      "epoch": 0.030256012412723042,
+      "grad_norm": 16.372045516967773,
+      "learning_rate": 1.5217391304347825e-06,
+      "loss": 6.3741,
+      "step": 39
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "grad_norm": 10.656290054321289,
+      "learning_rate": 1.562867215041128e-06,
+      "loss": 5.3981,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_Qnli-dev_cosine_accuracy": 0.6015625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9430716633796692,
+      "eval_Qnli-dev_cosine_ap": 0.5565875119399253,
+      "eval_Qnli-dev_cosine_f1": 0.6291834002677376,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.6778229475021362,
+      "eval_Qnli-dev_cosine_precision": 0.4598825831702544,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.548828125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 709.1221923828125,
+      "eval_Qnli-dev_dot_ap": 0.47478000038083695,
+      "eval_Qnli-dev_dot_f1": 0.6291834002677376,
+      "eval_Qnli-dev_dot_f1_threshold": 357.37890625,
+      "eval_Qnli-dev_dot_precision": 0.4598825831702544,
+      "eval_Qnli-dev_dot_recall": 0.9957627118644068,
+      "eval_Qnli-dev_euclidean_accuracy": 0.599609375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 8.403730392456055,
+      "eval_Qnli-dev_euclidean_ap": 0.5735439175412295,
+      "eval_Qnli-dev_euclidean_f1": 0.6324324324324324,
+      "eval_Qnli-dev_euclidean_f1_threshold": 17.377689361572266,
+      "eval_Qnli-dev_euclidean_precision": 0.4642857142857143,
+      "eval_Qnli-dev_euclidean_recall": 0.9915254237288136,
+      "eval_Qnli-dev_manhattan_accuracy": 0.625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 152.123046875,
+      "eval_Qnli-dev_manhattan_ap": 0.6110278070440811,
+      "eval_Qnli-dev_manhattan_f1": 0.6291834002677376,
+      "eval_Qnli-dev_manhattan_f1_threshold": 275.72344970703125,
+      "eval_Qnli-dev_manhattan_precision": 0.4598825831702544,
+      "eval_Qnli-dev_manhattan_recall": 0.9957627118644068,
+      "eval_Qnli-dev_max_accuracy": 0.625,
+      "eval_Qnli-dev_max_accuracy_threshold": 709.1221923828125,
+      "eval_Qnli-dev_max_ap": 0.6110278070440811,
+      "eval_Qnli-dev_max_f1": 0.6324324324324324,
+      "eval_Qnli-dev_max_f1_threshold": 357.37890625,
+      "eval_Qnli-dev_max_precision": 0.4642857142857143,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.666015625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9908591508865356,
+      "eval_allNLI-dev_cosine_ap": 0.3677090943014154,
+      "eval_allNLI-dev_cosine_f1": 0.5088235294117648,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8096739053726196,
+      "eval_allNLI-dev_cosine_precision": 0.34122287968441817,
+      "eval_allNLI-dev_cosine_recall": 1.0,
+      "eval_allNLI-dev_dot_accuracy": 0.66015625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 1065.9342041015625,
+      "eval_allNLI-dev_dot_ap": 0.33310258182405583,
+      "eval_allNLI-dev_dot_f1": 0.5081240768094535,
+      "eval_allNLI-dev_dot_f1_threshold": 600.9107666015625,
+      "eval_allNLI-dev_dot_precision": 0.3412698412698413,
+      "eval_allNLI-dev_dot_recall": 0.9942196531791907,
+      "eval_allNLI-dev_euclidean_accuracy": 0.666015625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.249004364013672,
+      "eval_allNLI-dev_euclidean_ap": 0.36405301502121135,
+      "eval_allNLI-dev_euclidean_f1": 0.5103857566765578,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.79437828063965,
+      "eval_allNLI-dev_euclidean_precision": 0.34331337325349304,
+      "eval_allNLI-dev_euclidean_recall": 0.9942196531791907,
+      "eval_allNLI-dev_manhattan_accuracy": 0.666015625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 71.94633483886719,
+      "eval_allNLI-dev_manhattan_ap": 0.3767657897123271,
+      "eval_allNLI-dev_manhattan_f1": 0.5073746312684366,
+      "eval_allNLI-dev_manhattan_f1_threshold": 236.0598602294922,
+      "eval_allNLI-dev_manhattan_precision": 0.3405940594059406,
+      "eval_allNLI-dev_manhattan_recall": 0.9942196531791907,
+      "eval_allNLI-dev_max_accuracy": 0.666015625,
+      "eval_allNLI-dev_max_accuracy_threshold": 1065.9342041015625,
+      "eval_allNLI-dev_max_ap": 0.3767657897123271,
+      "eval_allNLI-dev_max_f1": 0.5103857566765578,
+      "eval_allNLI-dev_max_f1_threshold": 600.9107666015625,
+      "eval_allNLI-dev_max_precision": 0.34331337325349304,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6110278070440811,
+      "eval_sts-test_pearson_cosine": 0.14752508082997276,
+      "eval_sts-test_pearson_dot": 0.13488717804142986,
+      "eval_sts-test_pearson_euclidean": 0.15185506014664446,
+      "eval_sts-test_pearson_manhattan": 0.14687917919770682,
+      "eval_sts-test_pearson_max": 0.15185506014664446,
+      "eval_sts-test_spearman_cosine": 0.20049724161017018,
+      "eval_sts-test_spearman_dot": 0.1353592298330889,
+      "eval_sts-test_spearman_euclidean": 0.17951910330520107,
+      "eval_sts-test_spearman_manhattan": 0.17376785019622396,
+      "eval_sts-test_spearman_max": 0.20049724161017018,
+      "eval_vitaminc-pairs_loss": 4.123702049255371,
+      "eval_vitaminc-pairs_runtime": 5.4736,
+      "eval_vitaminc-pairs_samples_per_second": 23.385,
+      "eval_vitaminc-pairs_steps_per_second": 0.365,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_negation-triplets_loss": 5.45994758605957,
+      "eval_negation-triplets_runtime": 1.0747,
+      "eval_negation-triplets_samples_per_second": 119.102,
+      "eval_negation-triplets_steps_per_second": 1.861,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_scitail-pairs-pos_loss": 2.758348226547241,
+      "eval_scitail-pairs-pos_runtime": 1.2143,
+      "eval_scitail-pairs-pos_samples_per_second": 105.41,
+      "eval_scitail-pairs-pos_steps_per_second": 1.647,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_scitail-pairs-qa_loss": 4.180817127227783,
+      "eval_scitail-pairs-qa_runtime": 0.766,
+      "eval_scitail-pairs-qa_samples_per_second": 167.105,
+      "eval_scitail-pairs-qa_steps_per_second": 2.611,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_xsum-pairs_loss": 5.874823093414307,
+      "eval_xsum-pairs_runtime": 3.924,
+      "eval_xsum-pairs_samples_per_second": 32.62,
+      "eval_xsum-pairs_steps_per_second": 0.51,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_sciq_pairs_loss": 0.6976423859596252,
+      "eval_sciq_pairs_runtime": 5.8211,
+      "eval_sciq_pairs_samples_per_second": 21.989,
+      "eval_sciq_pairs_steps_per_second": 0.344,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_qasc_pairs_loss": 7.423002243041992,
+      "eval_qasc_pairs_runtime": 0.9034,
+      "eval_qasc_pairs_samples_per_second": 141.693,
+      "eval_qasc_pairs_steps_per_second": 2.214,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_openbookqa_pairs_loss": 5.564730644226074,
+      "eval_openbookqa_pairs_runtime": 0.8569,
+      "eval_openbookqa_pairs_samples_per_second": 149.376,
+      "eval_openbookqa_pairs_steps_per_second": 2.334,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_msmarco_pairs_loss": 10.023717880249023,
+      "eval_msmarco_pairs_runtime": 1.9602,
+      "eval_msmarco_pairs_samples_per_second": 65.298,
+      "eval_msmarco_pairs_steps_per_second": 1.02,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_nq_pairs_loss": 8.906123161315918,
+      "eval_nq_pairs_runtime": 3.3287,
+      "eval_nq_pairs_samples_per_second": 38.453,
+      "eval_nq_pairs_steps_per_second": 0.601,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_trivia_pairs_loss": 6.981179714202881,
+      "eval_trivia_pairs_runtime": 4.4701,
+      "eval_trivia_pairs_samples_per_second": 28.635,
+      "eval_trivia_pairs_steps_per_second": 0.447,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_gooaq_pairs_loss": 7.387185096740723,
+      "eval_gooaq_pairs_runtime": 1.453,
+      "eval_gooaq_pairs_samples_per_second": 88.091,
+      "eval_gooaq_pairs_steps_per_second": 1.376,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_paws-pos_loss": 2.5774118900299072,
+      "eval_paws-pos_runtime": 1.0228,
+      "eval_paws-pos_samples_per_second": 125.149,
+      "eval_paws-pos_steps_per_second": 1.955,
+      "step": 40
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "eval_global_dataset_loss": 5.574436187744141,
+      "eval_global_dataset_runtime": 19.3888,
+      "eval_global_dataset_samples_per_second": 21.456,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 40
+    },
+    {
+      "epoch": 0.03180760279286268,
+      "grad_norm": 10.962615966796875,
+      "learning_rate": 1.6039952996474733e-06,
+      "loss": 5.3344,
+      "step": 41
+    },
+    {
+      "epoch": 0.032583397982932506,
+      "grad_norm": 14.594836235046387,
+      "learning_rate": 1.645123384253819e-06,
+      "loss": 6.1365,
+      "step": 42
+    },
+    {
+      "epoch": 0.03335919317300233,
+      "grad_norm": 11.795042037963867,
+      "learning_rate": 1.6862514688601645e-06,
+      "loss": 5.818,
+      "step": 43
+    },
+    {
+      "epoch": 0.03413498836307215,
+      "grad_norm": 12.332319259643555,
+      "learning_rate": 1.7273795534665098e-06,
+      "loss": 5.8133,
+      "step": 44
+    },
+    {
+      "epoch": 0.03491078355314197,
+      "grad_norm": 14.156538963317871,
+      "learning_rate": 1.7685076380728553e-06,
+      "loss": 5.9534,
+      "step": 45
+    },
+    {
+      "epoch": 0.035686578743211794,
+      "grad_norm": 11.36040210723877,
+      "learning_rate": 1.8096357226792008e-06,
+      "loss": 5.6302,
+      "step": 46
+    },
+    {
+      "epoch": 0.03646237393328161,
+      "grad_norm": 12.837271690368652,
+      "learning_rate": 1.8507638072855463e-06,
+      "loss": 5.9389,
+      "step": 47
+    },
+    {
+      "epoch": 0.037238169123351435,
+      "grad_norm": 13.556758880615234,
+      "learning_rate": 1.8918918918918918e-06,
+      "loss": 5.7893,
+      "step": 48
+    },
+    {
+      "epoch": 0.03801396431342126,
+      "grad_norm": 9.904508590698242,
+      "learning_rate": 1.933019976498237e-06,
+      "loss": 4.725,
+      "step": 49
+    },
+    {
+      "epoch": 0.038789759503491075,
+      "grad_norm": 13.982979774475098,
+      "learning_rate": 1.9741480611045828e-06,
+      "loss": 6.1877,
+      "step": 50
+    },
+    {
+      "epoch": 0.0395655546935609,
+      "grad_norm": 13.776800155639648,
+      "learning_rate": 2.015276145710928e-06,
+      "loss": 6.001,
+      "step": 51
+    },
+    {
+      "epoch": 0.04034134988363072,
+      "grad_norm": 21.959096908569336,
+      "learning_rate": 2.0564042303172738e-06,
+      "loss": 7.0504,
+      "step": 52
+    },
+    {
+      "epoch": 0.041117145073700546,
+      "grad_norm": 14.812617301940918,
+      "learning_rate": 2.097532314923619e-06,
+      "loss": 5.7706,
+      "step": 53
+    },
+    {
+      "epoch": 0.04189294026377036,
+      "grad_norm": 13.54603099822998,
+      "learning_rate": 2.1386603995299644e-06,
+      "loss": 6.3886,
+      "step": 54
+    },
+    {
+      "epoch": 0.04266873545384019,
+      "grad_norm": 11.500264167785645,
+      "learning_rate": 2.17978848413631e-06,
+      "loss": 6.0503,
+      "step": 55
+    },
+    {
+      "epoch": 0.04344453064391001,
+      "grad_norm": 7.893326759338379,
+      "learning_rate": 2.2209165687426554e-06,
+      "loss": 5.9136,
+      "step": 56
+    },
+    {
+      "epoch": 0.04422032583397983,
+      "grad_norm": 9.733207702636719,
+      "learning_rate": 2.2620446533490006e-06,
+      "loss": 5.6769,
+      "step": 57
+    },
+    {
+      "epoch": 0.04499612102404965,
+      "grad_norm": 12.644614219665527,
+      "learning_rate": 2.3031727379553464e-06,
+      "loss": 5.6418,
+      "step": 58
+    },
+    {
+      "epoch": 0.045771916214119475,
+      "grad_norm": 8.392154693603516,
+      "learning_rate": 2.344300822561692e-06,
+      "loss": 5.3225,
+      "step": 59
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "grad_norm": 6.115021228790283,
+      "learning_rate": 2.3854289071680374e-06,
+      "loss": 5.2977,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_Qnli-dev_cosine_accuracy": 0.59765625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9555542469024658,
+      "eval_Qnli-dev_cosine_ap": 0.5645675307780116,
+      "eval_Qnli-dev_cosine_f1": 0.6291834002677376,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7179017066955566,
+      "eval_Qnli-dev_cosine_precision": 0.4598825831702544,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.55078125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 727.389404296875,
+      "eval_Qnli-dev_dot_ap": 0.47243395169886154,
+      "eval_Qnli-dev_dot_f1": 0.629878869448183,
+      "eval_Qnli-dev_dot_f1_threshold": 461.4835205078125,
+      "eval_Qnli-dev_dot_precision": 0.46153846153846156,
+      "eval_Qnli-dev_dot_recall": 0.9915254237288136,
+      "eval_Qnli-dev_euclidean_accuracy": 0.599609375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 8.283002853393555,
+      "eval_Qnli-dev_euclidean_ap": 0.5826604533188524,
+      "eval_Qnli-dev_euclidean_f1": 0.6314363143631436,
+      "eval_Qnli-dev_euclidean_f1_threshold": 16.081790924072266,
+      "eval_Qnli-dev_euclidean_precision": 0.4641434262948207,
+      "eval_Qnli-dev_euclidean_recall": 0.9872881355932204,
+      "eval_Qnli-dev_manhattan_accuracy": 0.630859375,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 146.17575073242188,
+      "eval_Qnli-dev_manhattan_ap": 0.6168306227629134,
+      "eval_Qnli-dev_manhattan_f1": 0.6296296296296297,
+      "eval_Qnli-dev_manhattan_f1_threshold": 204.5635528564453,
+      "eval_Qnli-dev_manhattan_precision": 0.4742489270386266,
+      "eval_Qnli-dev_manhattan_recall": 0.9364406779661016,
+      "eval_Qnli-dev_max_accuracy": 0.630859375,
+      "eval_Qnli-dev_max_accuracy_threshold": 727.389404296875,
+      "eval_Qnli-dev_max_ap": 0.6168306227629134,
+      "eval_Qnli-dev_max_f1": 0.6314363143631436,
+      "eval_Qnli-dev_max_f1_threshold": 461.4835205078125,
+      "eval_Qnli-dev_max_precision": 0.4742489270386266,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.666015625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9849413633346558,
+      "eval_allNLI-dev_cosine_ap": 0.37307202329042877,
+      "eval_allNLI-dev_cosine_f1": 0.5104477611940298,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8623183965682983,
+      "eval_allNLI-dev_cosine_precision": 0.3440643863179074,
+      "eval_allNLI-dev_cosine_recall": 0.9884393063583815,
+      "eval_allNLI-dev_dot_accuracy": 0.66015625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 1062.3370361328125,
+      "eval_allNLI-dev_dot_ap": 0.3297699984610426,
+      "eval_allNLI-dev_dot_f1": 0.5088757396449705,
+      "eval_allNLI-dev_dot_f1_threshold": 622.9761962890625,
+      "eval_allNLI-dev_dot_precision": 0.341948310139165,
+      "eval_allNLI-dev_dot_recall": 0.9942196531791907,
+      "eval_allNLI-dev_euclidean_accuracy": 0.666015625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.127579689025879,
+      "eval_allNLI-dev_euclidean_ap": 0.36909730374104466,
+      "eval_allNLI-dev_euclidean_f1": 0.5113464447806354,
+      "eval_allNLI-dev_euclidean_f1_threshold": 14.735725402832031,
+      "eval_allNLI-dev_euclidean_precision": 0.3463114754098361,
+      "eval_allNLI-dev_euclidean_recall": 0.976878612716763,
+      "eval_allNLI-dev_manhattan_accuracy": 0.6640625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 54.84022521972656,
+      "eval_allNLI-dev_manhattan_ap": 0.3817636902039587,
+      "eval_allNLI-dev_manhattan_f1": 0.5103857566765578,
+      "eval_allNLI-dev_manhattan_f1_threshold": 225.62132263183594,
+      "eval_allNLI-dev_manhattan_precision": 0.34331337325349304,
+      "eval_allNLI-dev_manhattan_recall": 0.9942196531791907,
+      "eval_allNLI-dev_max_accuracy": 0.666015625,
+      "eval_allNLI-dev_max_accuracy_threshold": 1062.3370361328125,
+      "eval_allNLI-dev_max_ap": 0.3817636902039587,
+      "eval_allNLI-dev_max_f1": 0.5113464447806354,
+      "eval_allNLI-dev_max_f1_threshold": 622.9761962890625,
+      "eval_allNLI-dev_max_precision": 0.3463114754098361,
+      "eval_allNLI-dev_max_recall": 0.9942196531791907,
+      "eval_sequential_score": 0.6168306227629134,
+      "eval_sts-test_pearson_cosine": 0.15615571514608637,
+      "eval_sts-test_pearson_dot": 0.1565591340193878,
+      "eval_sts-test_pearson_euclidean": 0.15789005051166094,
+      "eval_sts-test_pearson_manhattan": 0.15049180567530787,
+      "eval_sts-test_pearson_max": 0.15789005051166094,
+      "eval_sts-test_spearman_cosine": 0.21185036811759986,
+      "eval_sts-test_spearman_dot": 0.15568507375201698,
+      "eval_sts-test_spearman_euclidean": 0.1835967487406626,
+      "eval_sts-test_spearman_manhattan": 0.17528196437414056,
+      "eval_sts-test_spearman_max": 0.21185036811759986,
+      "eval_vitaminc-pairs_loss": 3.992605209350586,
+      "eval_vitaminc-pairs_runtime": 5.4384,
+      "eval_vitaminc-pairs_samples_per_second": 23.536,
+      "eval_vitaminc-pairs_steps_per_second": 0.368,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_negation-triplets_loss": 5.40188455581665,
+      "eval_negation-triplets_runtime": 1.0636,
+      "eval_negation-triplets_samples_per_second": 120.348,
+      "eval_negation-triplets_steps_per_second": 1.88,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_scitail-pairs-pos_loss": 2.6962366104125977,
+      "eval_scitail-pairs-pos_runtime": 1.1899,
+      "eval_scitail-pairs-pos_samples_per_second": 107.569,
+      "eval_scitail-pairs-pos_steps_per_second": 1.681,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_scitail-pairs-qa_loss": 4.189697742462158,
+      "eval_scitail-pairs-qa_runtime": 0.7567,
+      "eval_scitail-pairs-qa_samples_per_second": 169.153,
+      "eval_scitail-pairs-qa_steps_per_second": 2.643,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_xsum-pairs_loss": 5.7225446701049805,
+      "eval_xsum-pairs_runtime": 3.9013,
+      "eval_xsum-pairs_samples_per_second": 32.809,
+      "eval_xsum-pairs_steps_per_second": 0.513,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_sciq_pairs_loss": 0.6581735610961914,
+      "eval_sciq_pairs_runtime": 5.7872,
+      "eval_sciq_pairs_samples_per_second": 22.118,
+      "eval_sciq_pairs_steps_per_second": 0.346,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_qasc_pairs_loss": 6.56049919128418,
+      "eval_qasc_pairs_runtime": 0.8954,
+      "eval_qasc_pairs_samples_per_second": 142.954,
+      "eval_qasc_pairs_steps_per_second": 2.234,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_openbookqa_pairs_loss": 5.458502292633057,
+      "eval_openbookqa_pairs_runtime": 0.8618,
+      "eval_openbookqa_pairs_samples_per_second": 148.532,
+      "eval_openbookqa_pairs_steps_per_second": 2.321,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_msmarco_pairs_loss": 8.660462379455566,
+      "eval_msmarco_pairs_runtime": 1.9838,
+      "eval_msmarco_pairs_samples_per_second": 64.521,
+      "eval_msmarco_pairs_steps_per_second": 1.008,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_nq_pairs_loss": 7.642305850982666,
+      "eval_nq_pairs_runtime": 3.3253,
+      "eval_nq_pairs_samples_per_second": 38.492,
+      "eval_nq_pairs_steps_per_second": 0.601,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_trivia_pairs_loss": 6.512197494506836,
+      "eval_trivia_pairs_runtime": 4.4671,
+      "eval_trivia_pairs_samples_per_second": 28.654,
+      "eval_trivia_pairs_steps_per_second": 0.448,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_gooaq_pairs_loss": 6.7339911460876465,
+      "eval_gooaq_pairs_runtime": 1.4577,
+      "eval_gooaq_pairs_samples_per_second": 87.81,
+      "eval_gooaq_pairs_steps_per_second": 1.372,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_paws-pos_loss": 2.735227584838867,
+      "eval_paws-pos_runtime": 1.0334,
+      "eval_paws-pos_samples_per_second": 123.864,
+      "eval_paws-pos_steps_per_second": 1.935,
+      "step": 60
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "eval_global_dataset_loss": 5.233480453491211,
+      "eval_global_dataset_runtime": 19.3785,
+      "eval_global_dataset_samples_per_second": 21.467,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 60
+    },
+    {
+      "epoch": 0.047323506594259115,
+      "grad_norm": 17.255535125732422,
+      "learning_rate": 2.426556991774383e-06,
+      "loss": 3.5938,
+      "step": 61
+    },
+    {
+      "epoch": 0.04809930178432894,
+      "grad_norm": 8.553607940673828,
+      "learning_rate": 2.4676850763807284e-06,
+      "loss": 6.1306,
+      "step": 62
+    },
+    {
+      "epoch": 0.048875096974398756,
+      "grad_norm": 29.715600967407227,
+      "learning_rate": 2.5088131609870737e-06,
+      "loss": 8.328,
+      "step": 63
+    },
+    {
+      "epoch": 0.04965089216446858,
+      "grad_norm": 11.584660530090332,
+      "learning_rate": 2.5499412455934194e-06,
+      "loss": 6.0765,
+      "step": 64
+    },
+    {
+      "epoch": 0.0504266873545384,
+      "grad_norm": 9.648879051208496,
+      "learning_rate": 2.5910693301997647e-06,
+      "loss": 4.808,
+      "step": 65
+    },
+    {
+      "epoch": 0.05120248254460822,
+      "grad_norm": 10.802507400512695,
+      "learning_rate": 2.63219741480611e-06,
+      "loss": 5.9628,
+      "step": 66
+    },
+    {
+      "epoch": 0.051978277734678044,
+      "grad_norm": 8.42625904083252,
+      "learning_rate": 2.6733254994124557e-06,
+      "loss": 5.197,
+      "step": 67
+    },
+    {
+      "epoch": 0.05275407292474787,
+      "grad_norm": 18.369491577148438,
+      "learning_rate": 2.7144535840188014e-06,
+      "loss": 6.7851,
+      "step": 68
+    },
+    {
+      "epoch": 0.05352986811481769,
+      "grad_norm": 16.186237335205078,
+      "learning_rate": 2.7555816686251467e-06,
+      "loss": 3.5248,
+      "step": 69
+    },
+    {
+      "epoch": 0.05430566330488751,
+      "grad_norm": 25.963136672973633,
+      "learning_rate": 2.796709753231492e-06,
+      "loss": 7.614,
+      "step": 70
+    },
+    {
+      "epoch": 0.05508145849495733,
+      "grad_norm": 10.458888053894043,
+      "learning_rate": 2.8378378378378377e-06,
+      "loss": 5.9919,
+      "step": 71
+    },
+    {
+      "epoch": 0.055857253685027156,
+      "grad_norm": 22.169750213623047,
+      "learning_rate": 2.878965922444183e-06,
+      "loss": 6.5741,
+      "step": 72
+    },
+    {
+      "epoch": 0.05663304887509697,
+      "grad_norm": 8.526782989501953,
+      "learning_rate": 2.9200940070505282e-06,
+      "loss": 5.5377,
+      "step": 73
+    },
+    {
+      "epoch": 0.057408844065166796,
+      "grad_norm": 17.269418716430664,
+      "learning_rate": 2.9612220916568744e-06,
+      "loss": 6.6046,
+      "step": 74
+    },
+    {
+      "epoch": 0.05818463925523662,
+      "grad_norm": 25.390901565551758,
+      "learning_rate": 3.0023501762632197e-06,
+      "loss": 6.6433,
+      "step": 75
+    },
+    {
+      "epoch": 0.05896043444530644,
+      "grad_norm": 9.188263893127441,
+      "learning_rate": 3.043478260869565e-06,
+      "loss": 5.347,
+      "step": 76
+    },
+    {
+      "epoch": 0.05973622963537626,
+      "grad_norm": 19.360456466674805,
+      "learning_rate": 3.0846063454759102e-06,
+      "loss": 6.3361,
+      "step": 77
+    },
+    {
+      "epoch": 0.060512024825446084,
+      "grad_norm": 16.59157371520996,
+      "learning_rate": 3.125734430082256e-06,
+      "loss": 6.672,
+      "step": 78
+    },
+    {
+      "epoch": 0.0612878200155159,
+      "grad_norm": 21.274341583251953,
+      "learning_rate": 3.1668625146886012e-06,
+      "loss": 7.2266,
+      "step": 79
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "grad_norm": 7.30432653427124,
+      "learning_rate": 3.2079905992949465e-06,
+      "loss": 5.2962,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_Qnli-dev_cosine_accuracy": 0.609375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9602965116500854,
+      "eval_Qnli-dev_cosine_ap": 0.5666608532248261,
+      "eval_Qnli-dev_cosine_f1": 0.6300268096514745,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8070105314254761,
+      "eval_Qnli-dev_cosine_precision": 0.46078431372549017,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.55078125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 752.1351318359375,
+      "eval_Qnli-dev_dot_ap": 0.47532727039762024,
+      "eval_Qnli-dev_dot_f1": 0.6307277628032345,
+      "eval_Qnli-dev_dot_f1_threshold": 492.98980712890625,
+      "eval_Qnli-dev_dot_precision": 0.4624505928853755,
+      "eval_Qnli-dev_dot_recall": 0.9915254237288136,
+      "eval_Qnli-dev_euclidean_accuracy": 0.609375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 7.72648811340332,
+      "eval_Qnli-dev_euclidean_ap": 0.5815079822504383,
+      "eval_Qnli-dev_euclidean_f1": 0.6300268096514745,
+      "eval_Qnli-dev_euclidean_f1_threshold": 16.727632522583008,
+      "eval_Qnli-dev_euclidean_precision": 0.46078431372549017,
+      "eval_Qnli-dev_euclidean_recall": 0.9957627118644068,
+      "eval_Qnli-dev_manhattan_accuracy": 0.625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 138.83407592773438,
+      "eval_Qnli-dev_manhattan_ap": 0.6149910998453122,
+      "eval_Qnli-dev_manhattan_f1": 0.6338028169014084,
+      "eval_Qnli-dev_manhattan_f1_threshold": 200.78419494628906,
+      "eval_Qnli-dev_manhattan_precision": 0.47468354430379744,
+      "eval_Qnli-dev_manhattan_recall": 0.9533898305084746,
+      "eval_Qnli-dev_max_accuracy": 0.625,
+      "eval_Qnli-dev_max_accuracy_threshold": 752.1351318359375,
+      "eval_Qnli-dev_max_ap": 0.6149910998453122,
+      "eval_Qnli-dev_max_f1": 0.6338028169014084,
+      "eval_Qnli-dev_max_f1_threshold": 492.98980712890625,
+      "eval_Qnli-dev_max_precision": 0.47468354430379744,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.66796875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9806065559387207,
+      "eval_allNLI-dev_cosine_ap": 0.3777500660247708,
+      "eval_allNLI-dev_cosine_f1": 0.5105105105105106,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8768577575683594,
+      "eval_allNLI-dev_cosine_precision": 0.3448275862068966,
+      "eval_allNLI-dev_cosine_recall": 0.9826589595375722,
+      "eval_allNLI-dev_dot_accuracy": 0.66015625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 1061.25537109375,
+      "eval_allNLI-dev_dot_ap": 0.32808659027780296,
+      "eval_allNLI-dev_dot_f1": 0.5103244837758112,
+      "eval_allNLI-dev_dot_f1_threshold": 628.6370239257812,
+      "eval_allNLI-dev_dot_precision": 0.3425742574257426,
+      "eval_allNLI-dev_dot_recall": 1.0,
+      "eval_allNLI-dev_euclidean_accuracy": 0.6640625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 2.8695411682128906,
+      "eval_allNLI-dev_euclidean_ap": 0.3719354048335051,
+      "eval_allNLI-dev_euclidean_f1": 0.5098634294385432,
+      "eval_allNLI-dev_euclidean_f1_threshold": 14.270792007446289,
+      "eval_allNLI-dev_euclidean_precision": 0.345679012345679,
+      "eval_allNLI-dev_euclidean_recall": 0.9710982658959537,
+      "eval_allNLI-dev_manhattan_accuracy": 0.6640625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 52.462921142578125,
+      "eval_allNLI-dev_manhattan_ap": 0.385436980166003,
+      "eval_allNLI-dev_manhattan_f1": 0.5121212121212121,
+      "eval_allNLI-dev_manhattan_f1_threshold": 210.9059295654297,
+      "eval_allNLI-dev_manhattan_precision": 0.3470225872689938,
+      "eval_allNLI-dev_manhattan_recall": 0.976878612716763,
+      "eval_allNLI-dev_max_accuracy": 0.66796875,
+      "eval_allNLI-dev_max_accuracy_threshold": 1061.25537109375,
+      "eval_allNLI-dev_max_ap": 0.385436980166003,
+      "eval_allNLI-dev_max_f1": 0.5121212121212121,
+      "eval_allNLI-dev_max_f1_threshold": 628.6370239257812,
+      "eval_allNLI-dev_max_precision": 0.3470225872689938,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6149910998453122,
+      "eval_sts-test_pearson_cosine": 0.16481809744211354,
+      "eval_sts-test_pearson_dot": 0.1687336929448065,
+      "eval_sts-test_pearson_euclidean": 0.16621273435380343,
+      "eval_sts-test_pearson_manhattan": 0.15708095069304162,
+      "eval_sts-test_pearson_max": 0.1687336929448065,
+      "eval_sts-test_spearman_cosine": 0.22067965773633885,
+      "eval_sts-test_spearman_dot": 0.16797399388640222,
+      "eval_sts-test_spearman_euclidean": 0.19304745437508256,
+      "eval_sts-test_spearman_manhattan": 0.18393475501795184,
+      "eval_sts-test_spearman_max": 0.22067965773633885,
+      "eval_vitaminc-pairs_loss": 3.9390244483947754,
+      "eval_vitaminc-pairs_runtime": 5.4765,
+      "eval_vitaminc-pairs_samples_per_second": 23.373,
+      "eval_vitaminc-pairs_steps_per_second": 0.365,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_negation-triplets_loss": 5.378707408905029,
+      "eval_negation-triplets_runtime": 1.0807,
+      "eval_negation-triplets_samples_per_second": 118.445,
+      "eval_negation-triplets_steps_per_second": 1.851,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_scitail-pairs-pos_loss": 2.6731348037719727,
+      "eval_scitail-pairs-pos_runtime": 1.2466,
+      "eval_scitail-pairs-pos_samples_per_second": 102.681,
+      "eval_scitail-pairs-pos_steps_per_second": 1.604,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_scitail-pairs-qa_loss": 4.19725227355957,
+      "eval_scitail-pairs-qa_runtime": 0.7667,
+      "eval_scitail-pairs-qa_samples_per_second": 166.945,
+      "eval_scitail-pairs-qa_steps_per_second": 2.609,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_xsum-pairs_loss": 5.497353553771973,
+      "eval_xsum-pairs_runtime": 3.9143,
+      "eval_xsum-pairs_samples_per_second": 32.701,
+      "eval_xsum-pairs_steps_per_second": 0.511,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_sciq_pairs_loss": 0.6366308927536011,
+      "eval_sciq_pairs_runtime": 5.8407,
+      "eval_sciq_pairs_samples_per_second": 21.915,
+      "eval_sciq_pairs_steps_per_second": 0.342,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_qasc_pairs_loss": 5.937916278839111,
+      "eval_qasc_pairs_runtime": 0.9043,
+      "eval_qasc_pairs_samples_per_second": 141.552,
+      "eval_qasc_pairs_steps_per_second": 2.212,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_openbookqa_pairs_loss": 5.383679389953613,
+      "eval_openbookqa_pairs_runtime": 0.8569,
+      "eval_openbookqa_pairs_samples_per_second": 149.38,
+      "eval_openbookqa_pairs_steps_per_second": 2.334,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_msmarco_pairs_loss": 7.691795825958252,
+      "eval_msmarco_pairs_runtime": 1.9631,
+      "eval_msmarco_pairs_samples_per_second": 65.203,
+      "eval_msmarco_pairs_steps_per_second": 1.019,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_nq_pairs_loss": 6.876422882080078,
+      "eval_nq_pairs_runtime": 3.3205,
+      "eval_nq_pairs_samples_per_second": 38.549,
+      "eval_nq_pairs_steps_per_second": 0.602,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_trivia_pairs_loss": 6.1294965744018555,
+      "eval_trivia_pairs_runtime": 4.4744,
+      "eval_trivia_pairs_samples_per_second": 28.607,
+      "eval_trivia_pairs_steps_per_second": 0.447,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_gooaq_pairs_loss": 6.205443859100342,
+      "eval_gooaq_pairs_runtime": 1.4558,
+      "eval_gooaq_pairs_samples_per_second": 87.923,
+      "eval_gooaq_pairs_steps_per_second": 1.374,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_paws-pos_loss": 2.7808141708374023,
+      "eval_paws-pos_runtime": 1.0141,
+      "eval_paws-pos_samples_per_second": 126.225,
+      "eval_paws-pos_steps_per_second": 1.972,
+      "step": 80
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "eval_global_dataset_loss": 5.0193328857421875,
+      "eval_global_dataset_runtime": 19.3707,
+      "eval_global_dataset_samples_per_second": 21.476,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 80
+    },
+    {
+      "epoch": 0.06283941039565555,
+      "grad_norm": 17.17608070373535,
+      "learning_rate": 3.2491186839012927e-06,
+      "loss": 6.1576,
+      "step": 81
+    },
+    {
+      "epoch": 0.06361520558572537,
+      "grad_norm": 22.93490982055664,
+      "learning_rate": 3.290246768507638e-06,
+      "loss": 7.3243,
+      "step": 82
+    },
+    {
+      "epoch": 0.0643910007757952,
+      "grad_norm": 7.930976867675781,
+      "learning_rate": 3.3313748531139832e-06,
+      "loss": 5.4146,
+      "step": 83
+    },
+    {
+      "epoch": 0.06516679596586501,
+      "grad_norm": 9.221315383911133,
+      "learning_rate": 3.372502937720329e-06,
+      "loss": 5.8422,
+      "step": 84
+    },
+    {
+      "epoch": 0.06594259115593483,
+      "grad_norm": 16.351512908935547,
+      "learning_rate": 3.4136310223266742e-06,
+      "loss": 3.7286,
+      "step": 85
+    },
+    {
+      "epoch": 0.06671838634600466,
+      "grad_norm": 6.853544235229492,
+      "learning_rate": 3.4547591069330195e-06,
+      "loss": 5.478,
+      "step": 86
+    },
+    {
+      "epoch": 0.06749418153607448,
+      "grad_norm": 7.5860700607299805,
+      "learning_rate": 3.495887191539365e-06,
+      "loss": 5.7321,
+      "step": 87
+    },
+    {
+      "epoch": 0.0682699767261443,
+      "grad_norm": 16.70624351501465,
+      "learning_rate": 3.5370152761457105e-06,
+      "loss": 3.6133,
+      "step": 88
+    },
+    {
+      "epoch": 0.06904577191621412,
+      "grad_norm": 20.827497482299805,
+      "learning_rate": 3.5781433607520563e-06,
+      "loss": 7.1329,
+      "step": 89
+    },
+    {
+      "epoch": 0.06982156710628394,
+      "grad_norm": 22.313615798950195,
+      "learning_rate": 3.6192714453584015e-06,
+      "loss": 7.1766,
+      "step": 90
+    },
+    {
+      "epoch": 0.07059736229635376,
+      "grad_norm": 12.5264892578125,
+      "learning_rate": 3.6603995299647473e-06,
+      "loss": 5.9805,
+      "step": 91
+    },
+    {
+      "epoch": 0.07137315748642359,
+      "grad_norm": 5.0581955909729,
+      "learning_rate": 3.7015276145710925e-06,
+      "loss": 5.797,
+      "step": 92
+    },
+    {
+      "epoch": 0.0721489526764934,
+      "grad_norm": 11.093539237976074,
+      "learning_rate": 3.742655699177438e-06,
+      "loss": 6.2104,
+      "step": 93
+    },
+    {
+      "epoch": 0.07292474786656322,
+      "grad_norm": 18.35002326965332,
+      "learning_rate": 3.7837837837837835e-06,
+      "loss": 6.7904,
+      "step": 94
+    },
+    {
+      "epoch": 0.07370054305663305,
+      "grad_norm": 4.62692928314209,
+      "learning_rate": 3.824911868390129e-06,
+      "loss": 5.6134,
+      "step": 95
+    },
+    {
+      "epoch": 0.07447633824670287,
+      "grad_norm": 5.769094944000244,
+      "learning_rate": 3.866039952996474e-06,
+      "loss": 5.5922,
+      "step": 96
+    },
+    {
+      "epoch": 0.07525213343677269,
+      "grad_norm": 6.5628461837768555,
+      "learning_rate": 3.907168037602819e-06,
+      "loss": 5.0351,
+      "step": 97
+    },
+    {
+      "epoch": 0.07602792862684252,
+      "grad_norm": 18.365070343017578,
+      "learning_rate": 3.9482961222091655e-06,
+      "loss": 7.0271,
+      "step": 98
+    },
+    {
+      "epoch": 0.07680372381691233,
+      "grad_norm": 12.308566093444824,
+      "learning_rate": 3.989424206815511e-06,
+      "loss": 6.2418,
+      "step": 99
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "grad_norm": 15.452849388122559,
+      "learning_rate": 4.030552291421856e-06,
+      "loss": 6.6114,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_Qnli-dev_cosine_accuracy": 0.6015625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9667960405349731,
+      "eval_Qnli-dev_cosine_ap": 0.5703877819293376,
+      "eval_Qnli-dev_cosine_f1": 0.6317204301075269,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8393322229385376,
+      "eval_Qnli-dev_cosine_precision": 0.4625984251968504,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.546875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 798.5418701171875,
+      "eval_Qnli-dev_dot_ap": 0.47597628614259624,
+      "eval_Qnli-dev_dot_f1": 0.6307277628032345,
+      "eval_Qnli-dev_dot_f1_threshold": 552.041015625,
+      "eval_Qnli-dev_dot_precision": 0.4624505928853755,
+      "eval_Qnli-dev_dot_recall": 0.9915254237288136,
+      "eval_Qnli-dev_euclidean_accuracy": 0.603515625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 7.820992469787598,
+      "eval_Qnli-dev_euclidean_ap": 0.5870442553034873,
+      "eval_Qnli-dev_euclidean_f1": 0.6300268096514745,
+      "eval_Qnli-dev_euclidean_f1_threshold": 16.982101440429688,
+      "eval_Qnli-dev_euclidean_precision": 0.46078431372549017,
+      "eval_Qnli-dev_euclidean_recall": 0.9957627118644068,
+      "eval_Qnli-dev_manhattan_accuracy": 0.62109375,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 138.17556762695312,
+      "eval_Qnli-dev_manhattan_ap": 0.613470320794373,
+      "eval_Qnli-dev_manhattan_f1": 0.6304347826086957,
+      "eval_Qnli-dev_manhattan_f1_threshold": 210.9376220703125,
+      "eval_Qnli-dev_manhattan_precision": 0.464,
+      "eval_Qnli-dev_manhattan_recall": 0.9830508474576272,
+      "eval_Qnli-dev_max_accuracy": 0.62109375,
+      "eval_Qnli-dev_max_accuracy_threshold": 798.5418701171875,
+      "eval_Qnli-dev_max_ap": 0.613470320794373,
+      "eval_Qnli-dev_max_f1": 0.6317204301075269,
+      "eval_Qnli-dev_max_f1_threshold": 552.041015625,
+      "eval_Qnli-dev_max_precision": 0.464,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.669921875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9819352626800537,
+      "eval_allNLI-dev_cosine_ap": 0.3832808530001383,
+      "eval_allNLI-dev_cosine_f1": 0.5082706766917294,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8935015201568604,
+      "eval_allNLI-dev_cosine_precision": 0.3434959349593496,
+      "eval_allNLI-dev_cosine_recall": 0.976878612716763,
+      "eval_allNLI-dev_dot_accuracy": 0.66015625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 1097.2392578125,
+      "eval_allNLI-dev_dot_ap": 0.32480014044358735,
+      "eval_allNLI-dev_dot_f1": 0.5095729013254787,
+      "eval_allNLI-dev_dot_f1_threshold": 669.5403442382812,
+      "eval_allNLI-dev_dot_precision": 0.34189723320158105,
+      "eval_allNLI-dev_dot_recall": 1.0,
+      "eval_allNLI-dev_euclidean_accuracy": 0.666015625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.188655376434326,
+      "eval_allNLI-dev_euclidean_ap": 0.3752436345647485,
+      "eval_allNLI-dev_euclidean_f1": 0.5112781954887218,
+      "eval_allNLI-dev_euclidean_f1_threshold": 14.686256408691406,
+      "eval_allNLI-dev_euclidean_precision": 0.34552845528455284,
+      "eval_allNLI-dev_euclidean_recall": 0.9826589595375722,
+      "eval_allNLI-dev_manhattan_accuracy": 0.66796875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 81.46263122558594,
+      "eval_allNLI-dev_manhattan_ap": 0.39261471479691196,
+      "eval_allNLI-dev_manhattan_f1": 0.513677811550152,
+      "eval_allNLI-dev_manhattan_f1_threshold": 204.65167236328125,
+      "eval_allNLI-dev_manhattan_precision": 0.34845360824742266,
+      "eval_allNLI-dev_manhattan_recall": 0.976878612716763,
+      "eval_allNLI-dev_max_accuracy": 0.669921875,
+      "eval_allNLI-dev_max_accuracy_threshold": 1097.2392578125,
+      "eval_allNLI-dev_max_ap": 0.39261471479691196,
+      "eval_allNLI-dev_max_f1": 0.513677811550152,
+      "eval_allNLI-dev_max_f1_threshold": 669.5403442382812,
+      "eval_allNLI-dev_max_precision": 0.34845360824742266,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.613470320794373,
+      "eval_sts-test_pearson_cosine": 0.1720525202149525,
+      "eval_sts-test_pearson_dot": 0.17715667152519826,
+      "eval_sts-test_pearson_euclidean": 0.17331039588378508,
+      "eval_sts-test_pearson_manhattan": 0.16267378650643669,
+      "eval_sts-test_pearson_max": 0.17715667152519826,
+      "eval_sts-test_spearman_cosine": 0.2278064303667034,
+      "eval_sts-test_spearman_dot": 0.17562364025486388,
+      "eval_sts-test_spearman_euclidean": 0.19856242082391304,
+      "eval_sts-test_spearman_manhattan": 0.1882869494420496,
+      "eval_sts-test_spearman_max": 0.2278064303667034,
+      "eval_vitaminc-pairs_loss": 3.920011281967163,
+      "eval_vitaminc-pairs_runtime": 5.4953,
+      "eval_vitaminc-pairs_samples_per_second": 23.293,
+      "eval_vitaminc-pairs_steps_per_second": 0.364,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_negation-triplets_loss": 5.354557991027832,
+      "eval_negation-triplets_runtime": 1.0836,
+      "eval_negation-triplets_samples_per_second": 118.12,
+      "eval_negation-triplets_steps_per_second": 1.846,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_scitail-pairs-pos_loss": 2.650925636291504,
+      "eval_scitail-pairs-pos_runtime": 1.2361,
+      "eval_scitail-pairs-pos_samples_per_second": 103.552,
+      "eval_scitail-pairs-pos_steps_per_second": 1.618,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_scitail-pairs-qa_loss": 4.260052680969238,
+      "eval_scitail-pairs-qa_runtime": 0.7647,
+      "eval_scitail-pairs-qa_samples_per_second": 167.375,
+      "eval_scitail-pairs-qa_steps_per_second": 2.615,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_xsum-pairs_loss": 5.345217227935791,
+      "eval_xsum-pairs_runtime": 3.917,
+      "eval_xsum-pairs_samples_per_second": 32.678,
+      "eval_xsum-pairs_steps_per_second": 0.511,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_sciq_pairs_loss": 0.6231272220611572,
+      "eval_sciq_pairs_runtime": 5.8736,
+      "eval_sciq_pairs_samples_per_second": 21.793,
+      "eval_sciq_pairs_steps_per_second": 0.341,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_qasc_pairs_loss": 5.399855136871338,
+      "eval_qasc_pairs_runtime": 0.9118,
+      "eval_qasc_pairs_samples_per_second": 140.388,
+      "eval_qasc_pairs_steps_per_second": 2.194,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_openbookqa_pairs_loss": 5.2847161293029785,
+      "eval_openbookqa_pairs_runtime": 0.8603,
+      "eval_openbookqa_pairs_samples_per_second": 148.787,
+      "eval_openbookqa_pairs_steps_per_second": 2.325,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_msmarco_pairs_loss": 6.665152072906494,
+      "eval_msmarco_pairs_runtime": 1.9686,
+      "eval_msmarco_pairs_samples_per_second": 65.021,
+      "eval_msmarco_pairs_steps_per_second": 1.016,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_nq_pairs_loss": 6.199686050415039,
+      "eval_nq_pairs_runtime": 3.3173,
+      "eval_nq_pairs_samples_per_second": 38.585,
+      "eval_nq_pairs_steps_per_second": 0.603,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_trivia_pairs_loss": 5.7401909828186035,
+      "eval_trivia_pairs_runtime": 4.4762,
+      "eval_trivia_pairs_samples_per_second": 28.595,
+      "eval_trivia_pairs_steps_per_second": 0.447,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_gooaq_pairs_loss": 5.792109489440918,
+      "eval_gooaq_pairs_runtime": 1.4527,
+      "eval_gooaq_pairs_samples_per_second": 88.113,
+      "eval_gooaq_pairs_steps_per_second": 1.377,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_paws-pos_loss": 2.8563427925109863,
+      "eval_paws-pos_runtime": 1.0232,
+      "eval_paws-pos_samples_per_second": 125.099,
+      "eval_paws-pos_steps_per_second": 1.955,
+      "step": 100
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_global_dataset_loss": 4.84829568862915,
+      "eval_global_dataset_runtime": 19.3977,
+      "eval_global_dataset_samples_per_second": 21.446,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 100
+    },
+    {
+      "epoch": 0.07835531419705198,
+      "grad_norm": 7.275325298309326,
+      "learning_rate": 4.071680376028202e-06,
+      "loss": 5.3294,
+      "step": 101
+    },
+    {
+      "epoch": 0.0791311093871218,
+      "grad_norm": 6.854203224182129,
+      "learning_rate": 4.1128084606345476e-06,
+      "loss": 5.0933,
+      "step": 102
+    },
+    {
+      "epoch": 0.07990690457719161,
+      "grad_norm": 7.766842365264893,
+      "learning_rate": 4.153936545240893e-06,
+      "loss": 4.9333,
+      "step": 103
+    },
+    {
+      "epoch": 0.08068269976726145,
+      "grad_norm": 7.3934478759765625,
+      "learning_rate": 4.195064629847238e-06,
+      "loss": 5.1327,
+      "step": 104
+    },
+    {
+      "epoch": 0.08145849495733126,
+      "grad_norm": 4.127883434295654,
+      "learning_rate": 4.236192714453583e-06,
+      "loss": 5.5859,
+      "step": 105
+    },
+    {
+      "epoch": 0.08223429014740109,
+      "grad_norm": 8.2522554397583,
+      "learning_rate": 4.277320799059929e-06,
+      "loss": 4.9192,
+      "step": 106
+    },
+    {
+      "epoch": 0.08301008533747091,
+      "grad_norm": 16.312501907348633,
+      "learning_rate": 4.318448883666274e-06,
+      "loss": 3.8092,
+      "step": 107
+    },
+    {
+      "epoch": 0.08378588052754073,
+      "grad_norm": 9.926016807556152,
+      "learning_rate": 4.35957696827262e-06,
+      "loss": 6.0247,
+      "step": 108
+    },
+    {
+      "epoch": 0.08456167571761056,
+      "grad_norm": 7.373856544494629,
+      "learning_rate": 4.400705052878966e-06,
+      "loss": 5.6878,
+      "step": 109
+    },
+    {
+      "epoch": 0.08533747090768037,
+      "grad_norm": 7.337434768676758,
+      "learning_rate": 4.441833137485311e-06,
+      "loss": 5.0953,
+      "step": 110
+    },
+    {
+      "epoch": 0.08611326609775019,
+      "grad_norm": 12.865263938903809,
+      "learning_rate": 4.482961222091657e-06,
+      "loss": 6.3449,
+      "step": 111
+    },
+    {
+      "epoch": 0.08688906128782002,
+      "grad_norm": 6.9256591796875,
+      "learning_rate": 4.524089306698001e-06,
+      "loss": 4.9109,
+      "step": 112
+    },
+    {
+      "epoch": 0.08766485647788984,
+      "grad_norm": 4.341299057006836,
+      "learning_rate": 4.5652173913043474e-06,
+      "loss": 5.6,
+      "step": 113
+    },
+    {
+      "epoch": 0.08844065166795965,
+      "grad_norm": 8.670815467834473,
+      "learning_rate": 4.606345475910693e-06,
+      "loss": 4.8765,
+      "step": 114
+    },
+    {
+      "epoch": 0.08921644685802949,
+      "grad_norm": 10.17563247680664,
+      "learning_rate": 4.647473560517038e-06,
+      "loss": 5.7062,
+      "step": 115
+    },
+    {
+      "epoch": 0.0899922420480993,
+      "grad_norm": 7.781512260437012,
+      "learning_rate": 4.688601645123384e-06,
+      "loss": 4.7221,
+      "step": 116
+    },
+    {
+      "epoch": 0.09076803723816912,
+      "grad_norm": 8.361579895019531,
+      "learning_rate": 4.7297297297297294e-06,
+      "loss": 4.8871,
+      "step": 117
+    },
+    {
+      "epoch": 0.09154383242823895,
+      "grad_norm": 9.360175132751465,
+      "learning_rate": 4.770857814336075e-06,
+      "loss": 4.7828,
+      "step": 118
+    },
+    {
+      "epoch": 0.09231962761830877,
+      "grad_norm": 4.893143177032471,
+      "learning_rate": 4.81198589894242e-06,
+      "loss": 5.5502,
+      "step": 119
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "grad_norm": 9.300333976745605,
+      "learning_rate": 4.853113983548766e-06,
+      "loss": 4.6165,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_Qnli-dev_cosine_accuracy": 0.6015625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9670753479003906,
+      "eval_Qnli-dev_cosine_ap": 0.5722899415341274,
+      "eval_Qnli-dev_cosine_f1": 0.6317204301075269,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.79914391040802,
+      "eval_Qnli-dev_cosine_precision": 0.4625984251968504,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.548828125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 770.8974609375,
+      "eval_Qnli-dev_dot_ap": 0.47908167649664435,
+      "eval_Qnli-dev_dot_f1": 0.6317204301075269,
+      "eval_Qnli-dev_dot_f1_threshold": 503.69384765625,
+      "eval_Qnli-dev_dot_precision": 0.4625984251968504,
+      "eval_Qnli-dev_dot_recall": 0.9957627118644068,
+      "eval_Qnli-dev_euclidean_accuracy": 0.603515625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 8.770221710205078,
+      "eval_Qnli-dev_euclidean_ap": 0.5921268027422615,
+      "eval_Qnli-dev_euclidean_f1": 0.6308724832214765,
+      "eval_Qnli-dev_euclidean_f1_threshold": 16.821897506713867,
+      "eval_Qnli-dev_euclidean_precision": 0.46168958742632615,
+      "eval_Qnli-dev_euclidean_recall": 0.9957627118644068,
+      "eval_Qnli-dev_manhattan_accuracy": 0.625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 144.48236083984375,
+      "eval_Qnli-dev_manhattan_ap": 0.6088252763279582,
+      "eval_Qnli-dev_manhattan_f1": 0.6329113924050632,
+      "eval_Qnli-dev_manhattan_f1_threshold": 209.13174438476562,
+      "eval_Qnli-dev_manhattan_precision": 0.47368421052631576,
+      "eval_Qnli-dev_manhattan_recall": 0.9533898305084746,
+      "eval_Qnli-dev_max_accuracy": 0.625,
+      "eval_Qnli-dev_max_accuracy_threshold": 770.8974609375,
+      "eval_Qnli-dev_max_ap": 0.6088252763279582,
+      "eval_Qnli-dev_max_f1": 0.6329113924050632,
+      "eval_Qnli-dev_max_f1_threshold": 503.69384765625,
+      "eval_Qnli-dev_max_precision": 0.47368421052631576,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.6796875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9780304431915283,
+      "eval_allNLI-dev_cosine_ap": 0.3992786539720722,
+      "eval_allNLI-dev_cosine_f1": 0.5098634294385432,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8867166042327881,
+      "eval_allNLI-dev_cosine_precision": 0.345679012345679,
+      "eval_allNLI-dev_cosine_recall": 0.9710982658959537,
+      "eval_allNLI-dev_dot_accuracy": 0.66015625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 993.18896484375,
+      "eval_allNLI-dev_dot_ap": 0.32359114658021904,
+      "eval_allNLI-dev_dot_f1": 0.5080763582966226,
+      "eval_allNLI-dev_dot_f1_threshold": 613.2178344726562,
+      "eval_allNLI-dev_dot_precision": 0.3405511811023622,
+      "eval_allNLI-dev_dot_recall": 1.0,
+      "eval_allNLI-dev_euclidean_accuracy": 0.671875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.144635200500488,
+      "eval_allNLI-dev_euclidean_ap": 0.3919696045318126,
+      "eval_allNLI-dev_euclidean_f1": 0.5120481927710844,
+      "eval_allNLI-dev_euclidean_f1_threshold": 14.77474594116211,
+      "eval_allNLI-dev_euclidean_precision": 0.34623217922606925,
+      "eval_allNLI-dev_euclidean_recall": 0.9826589595375722,
+      "eval_allNLI-dev_manhattan_accuracy": 0.66796875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 91.93115234375,
+      "eval_allNLI-dev_manhattan_ap": 0.404528316560946,
+      "eval_allNLI-dev_manhattan_f1": 0.5082212257100149,
+      "eval_allNLI-dev_manhattan_f1_threshold": 230.32806396484375,
+      "eval_allNLI-dev_manhattan_precision": 0.34274193548387094,
+      "eval_allNLI-dev_manhattan_recall": 0.9826589595375722,
+      "eval_allNLI-dev_max_accuracy": 0.6796875,
+      "eval_allNLI-dev_max_accuracy_threshold": 993.18896484375,
+      "eval_allNLI-dev_max_ap": 0.404528316560946,
+      "eval_allNLI-dev_max_f1": 0.5120481927710844,
+      "eval_allNLI-dev_max_f1_threshold": 613.2178344726562,
+      "eval_allNLI-dev_max_precision": 0.34623217922606925,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6088252763279582,
+      "eval_sts-test_pearson_cosine": 0.18225596814280462,
+      "eval_sts-test_pearson_dot": 0.19389246184304787,
+      "eval_sts-test_pearson_euclidean": 0.18398275691369742,
+      "eval_sts-test_pearson_manhattan": 0.17222751044724327,
+      "eval_sts-test_pearson_max": 0.19389246184304787,
+      "eval_sts-test_spearman_cosine": 0.23913221385342298,
+      "eval_sts-test_spearman_dot": 0.19319844513101708,
+      "eval_sts-test_spearman_euclidean": 0.2106999921922726,
+      "eval_sts-test_spearman_manhattan": 0.1993430170763632,
+      "eval_sts-test_spearman_max": 0.23913221385342298,
+      "eval_vitaminc-pairs_loss": 3.8911848068237305,
+      "eval_vitaminc-pairs_runtime": 5.47,
+      "eval_vitaminc-pairs_samples_per_second": 23.4,
+      "eval_vitaminc-pairs_steps_per_second": 0.366,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_negation-triplets_loss": 5.301531791687012,
+      "eval_negation-triplets_runtime": 1.0751,
+      "eval_negation-triplets_samples_per_second": 119.055,
+      "eval_negation-triplets_steps_per_second": 1.86,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_scitail-pairs-pos_loss": 2.535888910293579,
+      "eval_scitail-pairs-pos_runtime": 1.2447,
+      "eval_scitail-pairs-pos_samples_per_second": 102.834,
+      "eval_scitail-pairs-pos_steps_per_second": 1.607,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_scitail-pairs-qa_loss": 4.098081588745117,
+      "eval_scitail-pairs-qa_runtime": 0.7828,
+      "eval_scitail-pairs-qa_samples_per_second": 163.513,
+      "eval_scitail-pairs-qa_steps_per_second": 2.555,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_xsum-pairs_loss": 5.268539905548096,
+      "eval_xsum-pairs_runtime": 3.9213,
+      "eval_xsum-pairs_samples_per_second": 32.642,
+      "eval_xsum-pairs_steps_per_second": 0.51,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_sciq_pairs_loss": 0.607820987701416,
+      "eval_sciq_pairs_runtime": 5.8857,
+      "eval_sciq_pairs_samples_per_second": 21.748,
+      "eval_sciq_pairs_steps_per_second": 0.34,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_qasc_pairs_loss": 5.1934356689453125,
+      "eval_qasc_pairs_runtime": 0.9116,
+      "eval_qasc_pairs_samples_per_second": 140.413,
+      "eval_qasc_pairs_steps_per_second": 2.194,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_openbookqa_pairs_loss": 5.243656158447266,
+      "eval_openbookqa_pairs_runtime": 0.8604,
+      "eval_openbookqa_pairs_samples_per_second": 148.766,
+      "eval_openbookqa_pairs_steps_per_second": 2.324,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_msmarco_pairs_loss": 6.208409309387207,
+      "eval_msmarco_pairs_runtime": 1.9688,
+      "eval_msmarco_pairs_samples_per_second": 65.013,
+      "eval_msmarco_pairs_steps_per_second": 1.016,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_nq_pairs_loss": 5.938248157501221,
+      "eval_nq_pairs_runtime": 3.3259,
+      "eval_nq_pairs_samples_per_second": 38.486,
+      "eval_nq_pairs_steps_per_second": 0.601,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_trivia_pairs_loss": 5.63157844543457,
+      "eval_trivia_pairs_runtime": 4.4797,
+      "eval_trivia_pairs_samples_per_second": 28.574,
+      "eval_trivia_pairs_steps_per_second": 0.446,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_gooaq_pairs_loss": 5.568259239196777,
+      "eval_gooaq_pairs_runtime": 1.4576,
+      "eval_gooaq_pairs_samples_per_second": 87.818,
+      "eval_gooaq_pairs_steps_per_second": 1.372,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_paws-pos_loss": 2.3947365283966064,
+      "eval_paws-pos_runtime": 1.0203,
+      "eval_paws-pos_samples_per_second": 125.448,
+      "eval_paws-pos_steps_per_second": 1.96,
+      "step": 120
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "eval_global_dataset_loss": 4.51961088180542,
+      "eval_global_dataset_runtime": 19.3948,
+      "eval_global_dataset_samples_per_second": 21.449,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 120
+    },
+    {
+      "epoch": 0.09387121799844841,
+      "grad_norm": 8.517334938049316,
+      "learning_rate": 4.894242068155111e-06,
+      "loss": 4.892,
+      "step": 121
+    },
+    {
+      "epoch": 0.09464701318851823,
+      "grad_norm": 11.708131790161133,
+      "learning_rate": 4.935370152761457e-06,
+      "loss": 4.6732,
+      "step": 122
+    },
+    {
+      "epoch": 0.09542280837858805,
+      "grad_norm": 8.222532272338867,
+      "learning_rate": 4.976498237367803e-06,
+      "loss": 5.4697,
+      "step": 123
+    },
+    {
+      "epoch": 0.09619860356865788,
+      "grad_norm": 9.485191345214844,
+      "learning_rate": 5.017626321974147e-06,
+      "loss": 4.777,
+      "step": 124
+    },
+    {
+      "epoch": 0.0969743987587277,
+      "grad_norm": 19.364856719970703,
+      "learning_rate": 5.0587544065804934e-06,
+      "loss": 6.3552,
+      "step": 125
+    },
+    {
+      "epoch": 0.09775019394879751,
+      "grad_norm": 9.71495246887207,
+      "learning_rate": 5.099882491186839e-06,
+      "loss": 4.8758,
+      "step": 126
+    },
+    {
+      "epoch": 0.09852598913886734,
+      "grad_norm": 23.59428596496582,
+      "learning_rate": 5.141010575793184e-06,
+      "loss": 3.029,
+      "step": 127
+    },
+    {
+      "epoch": 0.09930178432893716,
+      "grad_norm": 15.757760047912598,
+      "learning_rate": 5.182138660399529e-06,
+      "loss": 5.7908,
+      "step": 128
+    },
+    {
+      "epoch": 0.10007757951900698,
+      "grad_norm": 20.3161678314209,
+      "learning_rate": 5.2232667450058754e-06,
+      "loss": 6.3179,
+      "step": 129
+    },
+    {
+      "epoch": 0.1008533747090768,
+      "grad_norm": 13.285233497619629,
+      "learning_rate": 5.26439482961222e-06,
+      "loss": 5.7515,
+      "step": 130
+    },
+    {
+      "epoch": 0.10162916989914662,
+      "grad_norm": 11.527393341064453,
+      "learning_rate": 5.305522914218566e-06,
+      "loss": 4.4534,
+      "step": 131
+    },
+    {
+      "epoch": 0.10240496508921644,
+      "grad_norm": 10.770364761352539,
+      "learning_rate": 5.346650998824911e-06,
+      "loss": 4.5598,
+      "step": 132
+    },
+    {
+      "epoch": 0.10318076027928627,
+      "grad_norm": 13.983633995056152,
+      "learning_rate": 5.387779083431257e-06,
+      "loss": 4.3048,
+      "step": 133
+    },
+    {
+      "epoch": 0.10395655546935609,
+      "grad_norm": 24.10936164855957,
+      "learning_rate": 5.428907168037603e-06,
+      "loss": 2.7566,
+      "step": 134
+    },
+    {
+      "epoch": 0.10473235065942592,
+      "grad_norm": 10.636360168457031,
+      "learning_rate": 5.470035252643947e-06,
+      "loss": 4.3254,
+      "step": 135
+    },
+    {
+      "epoch": 0.10550814584949574,
+      "grad_norm": 10.882925987243652,
+      "learning_rate": 5.511163337250293e-06,
+      "loss": 4.4835,
+      "step": 136
+    },
+    {
+      "epoch": 0.10628394103956555,
+      "grad_norm": 23.658288955688477,
+      "learning_rate": 5.5522914218566394e-06,
+      "loss": 4.4302,
+      "step": 137
+    },
+    {
+      "epoch": 0.10705973622963538,
+      "grad_norm": 11.732893943786621,
+      "learning_rate": 5.593419506462984e-06,
+      "loss": 4.2167,
+      "step": 138
+    },
+    {
+      "epoch": 0.1078355314197052,
+      "grad_norm": 15.095183372497559,
+      "learning_rate": 5.63454759106933e-06,
+      "loss": 4.3158,
+      "step": 139
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "grad_norm": 19.106853485107422,
+      "learning_rate": 5.675675675675675e-06,
+      "loss": 5.9198,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_Qnli-dev_cosine_accuracy": 0.6015625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.932540237903595,
+      "eval_Qnli-dev_cosine_ap": 0.590162641547892,
+      "eval_Qnli-dev_cosine_f1": 0.6358381502890174,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8430850505828857,
+      "eval_Qnli-dev_cosine_precision": 0.4824561403508772,
+      "eval_Qnli-dev_cosine_recall": 0.9322033898305084,
+      "eval_Qnli-dev_dot_accuracy": 0.568359375,
+      "eval_Qnli-dev_dot_accuracy_threshold": 651.802734375,
+      "eval_Qnli-dev_dot_ap": 0.4990308476722506,
+      "eval_Qnli-dev_dot_f1": 0.6318607764390897,
+      "eval_Qnli-dev_dot_f1_threshold": 359.81072998046875,
+      "eval_Qnli-dev_dot_precision": 0.461839530332681,
+      "eval_Qnli-dev_dot_recall": 1.0,
+      "eval_Qnli-dev_euclidean_accuracy": 0.607421875,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 10.347358703613281,
+      "eval_Qnli-dev_euclidean_ap": 0.5989545722402487,
+      "eval_Qnli-dev_euclidean_f1": 0.6300268096514745,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.583602905273438,
+      "eval_Qnli-dev_euclidean_precision": 0.46078431372549017,
+      "eval_Qnli-dev_euclidean_recall": 0.9957627118644068,
+      "eval_Qnli-dev_manhattan_accuracy": 0.62109375,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 161.458251953125,
+      "eval_Qnli-dev_manhattan_ap": 0.6134087495053078,
+      "eval_Qnli-dev_manhattan_f1": 0.6406685236768802,
+      "eval_Qnli-dev_manhattan_f1_threshold": 260.0387268066406,
+      "eval_Qnli-dev_manhattan_precision": 0.47717842323651455,
+      "eval_Qnli-dev_manhattan_recall": 0.9745762711864406,
+      "eval_Qnli-dev_max_accuracy": 0.62109375,
+      "eval_Qnli-dev_max_accuracy_threshold": 651.802734375,
+      "eval_Qnli-dev_max_ap": 0.6134087495053078,
+      "eval_Qnli-dev_max_f1": 0.6406685236768802,
+      "eval_Qnli-dev_max_f1_threshold": 359.81072998046875,
+      "eval_Qnli-dev_max_precision": 0.4824561403508772,
+      "eval_Qnli-dev_max_recall": 1.0,
+      "eval_allNLI-dev_cosine_accuracy": 0.68359375,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.971808135509491,
+      "eval_allNLI-dev_cosine_ap": 0.41916615465917384,
+      "eval_allNLI-dev_cosine_f1": 0.5088,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8694682121276855,
+      "eval_allNLI-dev_cosine_precision": 0.35176991150442477,
+      "eval_allNLI-dev_cosine_recall": 0.9190751445086706,
+      "eval_allNLI-dev_dot_accuracy": 0.6640625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 875.8065185546875,
+      "eval_allNLI-dev_dot_ap": 0.33345963244863186,
+      "eval_allNLI-dev_dot_f1": 0.5081723625557207,
+      "eval_allNLI-dev_dot_f1_threshold": 514.4766845703125,
+      "eval_allNLI-dev_dot_precision": 0.342,
+      "eval_allNLI-dev_dot_recall": 0.9884393063583815,
+      "eval_allNLI-dev_euclidean_accuracy": 0.673828125,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.979422092437744,
+      "eval_allNLI-dev_euclidean_ap": 0.41272090015066604,
+      "eval_allNLI-dev_euclidean_f1": 0.5102362204724409,
+      "eval_allNLI-dev_euclidean_f1_threshold": 14.31092643737793,
+      "eval_allNLI-dev_euclidean_precision": 0.35064935064935066,
+      "eval_allNLI-dev_euclidean_recall": 0.9364161849710982,
+      "eval_allNLI-dev_manhattan_accuracy": 0.673828125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 113.5325698852539,
+      "eval_allNLI-dev_manhattan_ap": 0.4170134796589544,
+      "eval_allNLI-dev_manhattan_f1": 0.5065885797950219,
+      "eval_allNLI-dev_manhattan_f1_threshold": 311.357177734375,
+      "eval_allNLI-dev_manhattan_precision": 0.3392156862745098,
+      "eval_allNLI-dev_manhattan_recall": 1.0,
+      "eval_allNLI-dev_max_accuracy": 0.68359375,
+      "eval_allNLI-dev_max_accuracy_threshold": 875.8065185546875,
+      "eval_allNLI-dev_max_ap": 0.41916615465917384,
+      "eval_allNLI-dev_max_f1": 0.5102362204724409,
+      "eval_allNLI-dev_max_f1_threshold": 514.4766845703125,
+      "eval_allNLI-dev_max_precision": 0.35176991150442477,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6134087495053078,
+      "eval_sts-test_pearson_cosine": 0.19606202486057506,
+      "eval_sts-test_pearson_dot": 0.20910750938672273,
+      "eval_sts-test_pearson_euclidean": 0.20204170742350583,
+      "eval_sts-test_pearson_manhattan": 0.19980066844086258,
+      "eval_sts-test_pearson_max": 0.20910750938672273,
+      "eval_sts-test_spearman_cosine": 0.2570204565643615,
+      "eval_sts-test_spearman_dot": 0.2142791085758339,
+      "eval_sts-test_spearman_euclidean": 0.22658230436260302,
+      "eval_sts-test_spearman_manhattan": 0.2257667126237448,
+      "eval_sts-test_spearman_max": 0.2570204565643615,
+      "eval_vitaminc-pairs_loss": 3.860930919647217,
+      "eval_vitaminc-pairs_runtime": 5.4616,
+      "eval_vitaminc-pairs_samples_per_second": 23.437,
+      "eval_vitaminc-pairs_steps_per_second": 0.366,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_negation-triplets_loss": 5.22545051574707,
+      "eval_negation-triplets_runtime": 1.061,
+      "eval_negation-triplets_samples_per_second": 120.643,
+      "eval_negation-triplets_steps_per_second": 1.885,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_scitail-pairs-pos_loss": 2.2156143188476562,
+      "eval_scitail-pairs-pos_runtime": 1.2311,
+      "eval_scitail-pairs-pos_samples_per_second": 103.974,
+      "eval_scitail-pairs-pos_steps_per_second": 1.625,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_scitail-pairs-qa_loss": 3.2973973751068115,
+      "eval_scitail-pairs-qa_runtime": 0.7709,
+      "eval_scitail-pairs-qa_samples_per_second": 166.038,
+      "eval_scitail-pairs-qa_steps_per_second": 2.594,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_xsum-pairs_loss": 5.160762310028076,
+      "eval_xsum-pairs_runtime": 3.9073,
+      "eval_xsum-pairs_samples_per_second": 32.759,
+      "eval_xsum-pairs_steps_per_second": 0.512,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_sciq_pairs_loss": 0.6155482530593872,
+      "eval_sciq_pairs_runtime": 5.8124,
+      "eval_sciq_pairs_samples_per_second": 22.022,
+      "eval_sciq_pairs_steps_per_second": 0.344,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_qasc_pairs_loss": 5.0954060554504395,
+      "eval_qasc_pairs_runtime": 0.9074,
+      "eval_qasc_pairs_samples_per_second": 141.069,
+      "eval_qasc_pairs_steps_per_second": 2.204,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_openbookqa_pairs_loss": 5.480427265167236,
+      "eval_openbookqa_pairs_runtime": 0.8517,
+      "eval_openbookqa_pairs_samples_per_second": 150.279,
+      "eval_openbookqa_pairs_steps_per_second": 2.348,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_msmarco_pairs_loss": 6.218542575836182,
+      "eval_msmarco_pairs_runtime": 1.9635,
+      "eval_msmarco_pairs_samples_per_second": 65.189,
+      "eval_msmarco_pairs_steps_per_second": 1.019,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_nq_pairs_loss": 5.9674296379089355,
+      "eval_nq_pairs_runtime": 3.3526,
+      "eval_nq_pairs_samples_per_second": 38.179,
+      "eval_nq_pairs_steps_per_second": 0.597,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_trivia_pairs_loss": 6.06894063949585,
+      "eval_trivia_pairs_runtime": 4.4755,
+      "eval_trivia_pairs_samples_per_second": 28.6,
+      "eval_trivia_pairs_steps_per_second": 0.447,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_gooaq_pairs_loss": 5.648138523101807,
+      "eval_gooaq_pairs_runtime": 1.4628,
+      "eval_gooaq_pairs_samples_per_second": 87.504,
+      "eval_gooaq_pairs_steps_per_second": 1.367,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_paws-pos_loss": 1.3320106267929077,
+      "eval_paws-pos_runtime": 1.0268,
+      "eval_paws-pos_samples_per_second": 124.656,
+      "eval_paws-pos_steps_per_second": 1.948,
+      "step": 140
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "eval_global_dataset_loss": 4.069947242736816,
+      "eval_global_dataset_runtime": 19.4022,
+      "eval_global_dataset_samples_per_second": 21.441,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 140
+    },
+    {
+      "epoch": 0.10938712179984485,
+      "grad_norm": 17.907608032226562,
+      "learning_rate": 5.716803760282021e-06,
+      "loss": 5.8619,
+      "step": 141
+    },
+    {
+      "epoch": 0.11016291698991466,
+      "grad_norm": 12.681406021118164,
+      "learning_rate": 5.757931844888366e-06,
+      "loss": 4.42,
+      "step": 142
+    },
+    {
+      "epoch": 0.11093871217998448,
+      "grad_norm": 29.748260498046875,
+      "learning_rate": 5.799059929494712e-06,
+      "loss": 6.5458,
+      "step": 143
+    },
+    {
+      "epoch": 0.11171450737005431,
+      "grad_norm": 17.36277198791504,
+      "learning_rate": 5.8401880141010565e-06,
+      "loss": 3.7463,
+      "step": 144
+    },
+    {
+      "epoch": 0.11249030256012413,
+      "grad_norm": 15.745108604431152,
+      "learning_rate": 5.881316098707403e-06,
+      "loss": 4.8236,
+      "step": 145
+    },
+    {
+      "epoch": 0.11326609775019394,
+      "grad_norm": 17.49993896484375,
+      "learning_rate": 5.922444183313749e-06,
+      "loss": 5.9082,
+      "step": 146
+    },
+    {
+      "epoch": 0.11404189294026378,
+      "grad_norm": 11.602448463439941,
+      "learning_rate": 5.963572267920093e-06,
+      "loss": 3.9276,
+      "step": 147
+    },
+    {
+      "epoch": 0.11481768813033359,
+      "grad_norm": 14.728039741516113,
+      "learning_rate": 6.004700352526439e-06,
+      "loss": 4.3073,
+      "step": 148
+    },
+    {
+      "epoch": 0.11559348332040341,
+      "grad_norm": 12.118491172790527,
+      "learning_rate": 6.045828437132785e-06,
+      "loss": 4.06,
+      "step": 149
+    },
+    {
+      "epoch": 0.11636927851047324,
+      "grad_norm": 14.151768684387207,
+      "learning_rate": 6.08695652173913e-06,
+      "loss": 4.7036,
+      "step": 150
+    },
+    {
+      "epoch": 0.11714507370054306,
+      "grad_norm": 12.606175422668457,
+      "learning_rate": 6.128084606345475e-06,
+      "loss": 3.9509,
+      "step": 151
+    },
+    {
+      "epoch": 0.11792086889061287,
+      "grad_norm": 38.992183685302734,
+      "learning_rate": 6.1692126909518205e-06,
+      "loss": 6.3894,
+      "step": 152
+    },
+    {
+      "epoch": 0.1186966640806827,
+      "grad_norm": 13.61286449432373,
+      "learning_rate": 6.210340775558167e-06,
+      "loss": 4.4063,
+      "step": 153
+    },
+    {
+      "epoch": 0.11947245927075252,
+      "grad_norm": 17.750974655151367,
+      "learning_rate": 6.251468860164512e-06,
+      "loss": 3.6723,
+      "step": 154
+    },
+    {
+      "epoch": 0.12024825446082234,
+      "grad_norm": 17.65484046936035,
+      "learning_rate": 6.292596944770857e-06,
+      "loss": 5.462,
+      "step": 155
+    },
+    {
+      "epoch": 0.12102404965089217,
+      "grad_norm": 12.64035415649414,
+      "learning_rate": 6.3337250293772025e-06,
+      "loss": 4.579,
+      "step": 156
+    },
+    {
+      "epoch": 0.12179984484096198,
+      "grad_norm": 11.979784965515137,
+      "learning_rate": 6.374853113983549e-06,
+      "loss": 3.9228,
+      "step": 157
+    },
+    {
+      "epoch": 0.1225756400310318,
+      "grad_norm": 12.76739501953125,
+      "learning_rate": 6.415981198589893e-06,
+      "loss": 3.6178,
+      "step": 158
+    },
+    {
+      "epoch": 0.12335143522110163,
+      "grad_norm": 27.8046817779541,
+      "learning_rate": 6.457109283196239e-06,
+      "loss": 5.686,
+      "step": 159
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "grad_norm": 12.14214038848877,
+      "learning_rate": 6.498237367802585e-06,
+      "loss": 3.5491,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_Qnli-dev_cosine_accuracy": 0.611328125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9235547780990601,
+      "eval_Qnli-dev_cosine_ap": 0.5831965528167927,
+      "eval_Qnli-dev_cosine_f1": 0.6318758815232722,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8174055218696594,
+      "eval_Qnli-dev_cosine_precision": 0.47357293868921774,
+      "eval_Qnli-dev_cosine_recall": 0.9491525423728814,
+      "eval_Qnli-dev_dot_accuracy": 0.564453125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 650.8389892578125,
+      "eval_Qnli-dev_dot_ap": 0.5067183783295084,
+      "eval_Qnli-dev_dot_f1": 0.6327077747989276,
+      "eval_Qnli-dev_dot_f1_threshold": 399.63836669921875,
+      "eval_Qnli-dev_dot_precision": 0.4627450980392157,
+      "eval_Qnli-dev_dot_recall": 1.0,
+      "eval_Qnli-dev_euclidean_accuracy": 0.6171875,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 10.054267883300781,
+      "eval_Qnli-dev_euclidean_ap": 0.5948450213638221,
+      "eval_Qnli-dev_euclidean_f1": 0.6318607764390897,
+      "eval_Qnli-dev_euclidean_f1_threshold": 20.041950225830078,
+      "eval_Qnli-dev_euclidean_precision": 0.461839530332681,
+      "eval_Qnli-dev_euclidean_recall": 1.0,
+      "eval_Qnli-dev_manhattan_accuracy": 0.625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 173.52857971191406,
+      "eval_Qnli-dev_manhattan_ap": 0.6122239571226092,
+      "eval_Qnli-dev_manhattan_f1": 0.6353591160220995,
+      "eval_Qnli-dev_manhattan_f1_threshold": 274.06744384765625,
+      "eval_Qnli-dev_manhattan_precision": 0.4713114754098361,
+      "eval_Qnli-dev_manhattan_recall": 0.9745762711864406,
+      "eval_Qnli-dev_max_accuracy": 0.625,
+      "eval_Qnli-dev_max_accuracy_threshold": 650.8389892578125,
+      "eval_Qnli-dev_max_ap": 0.6122239571226092,
+      "eval_Qnli-dev_max_f1": 0.6353591160220995,
+      "eval_Qnli-dev_max_f1_threshold": 399.63836669921875,
+      "eval_Qnli-dev_max_precision": 0.47357293868921774,
+      "eval_Qnli-dev_max_recall": 1.0,
+      "eval_allNLI-dev_cosine_accuracy": 0.6796875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9759693145751953,
+      "eval_allNLI-dev_cosine_ap": 0.42359590129311003,
+      "eval_allNLI-dev_cosine_f1": 0.5105740181268882,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8282982110977173,
+      "eval_allNLI-dev_cosine_precision": 0.3456032719836401,
+      "eval_allNLI-dev_cosine_recall": 0.976878612716763,
+      "eval_allNLI-dev_dot_accuracy": 0.66796875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 799.5001831054688,
+      "eval_allNLI-dev_dot_ap": 0.3510979636740506,
+      "eval_allNLI-dev_dot_f1": 0.5089820359281437,
+      "eval_allNLI-dev_dot_f1_threshold": 532.1497802734375,
+      "eval_allNLI-dev_dot_precision": 0.3434343434343434,
+      "eval_allNLI-dev_dot_recall": 0.9826589595375722,
+      "eval_allNLI-dev_euclidean_accuracy": 0.673828125,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.110721588134766,
+      "eval_allNLI-dev_euclidean_ap": 0.4178547437007212,
+      "eval_allNLI-dev_euclidean_f1": 0.519134775374376,
+      "eval_allNLI-dev_euclidean_f1_threshold": 13.405746459960938,
+      "eval_allNLI-dev_euclidean_precision": 0.3644859813084112,
+      "eval_allNLI-dev_euclidean_recall": 0.9017341040462428,
+      "eval_allNLI-dev_manhattan_accuracy": 0.67578125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 103.17061614990234,
+      "eval_allNLI-dev_manhattan_ap": 0.4272300717524159,
+      "eval_allNLI-dev_manhattan_f1": 0.5107033639143731,
+      "eval_allNLI-dev_manhattan_f1_threshold": 245.95425415039062,
+      "eval_allNLI-dev_manhattan_precision": 0.3471933471933472,
+      "eval_allNLI-dev_manhattan_recall": 0.9653179190751445,
+      "eval_allNLI-dev_max_accuracy": 0.6796875,
+      "eval_allNLI-dev_max_accuracy_threshold": 799.5001831054688,
+      "eval_allNLI-dev_max_ap": 0.4272300717524159,
+      "eval_allNLI-dev_max_f1": 0.519134775374376,
+      "eval_allNLI-dev_max_f1_threshold": 532.1497802734375,
+      "eval_allNLI-dev_max_precision": 0.3644859813084112,
+      "eval_allNLI-dev_max_recall": 0.9826589595375722,
+      "eval_sequential_score": 0.6122239571226092,
+      "eval_sts-test_pearson_cosine": 0.1983814843936258,
+      "eval_sts-test_pearson_dot": 0.21567697770934743,
+      "eval_sts-test_pearson_euclidean": 0.2076928261154226,
+      "eval_sts-test_pearson_manhattan": 0.2134685079189917,
+      "eval_sts-test_pearson_max": 0.21567697770934743,
+      "eval_sts-test_spearman_cosine": 0.2667653122399071,
+      "eval_sts-test_spearman_dot": 0.21757047434864857,
+      "eval_sts-test_spearman_euclidean": 0.2401902679264998,
+      "eval_sts-test_spearman_manhattan": 0.24235085882647658,
+      "eval_sts-test_spearman_max": 0.2667653122399071,
+      "eval_vitaminc-pairs_loss": 3.85381817817688,
+      "eval_vitaminc-pairs_runtime": 5.4462,
+      "eval_vitaminc-pairs_samples_per_second": 23.503,
+      "eval_vitaminc-pairs_steps_per_second": 0.367,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_negation-triplets_loss": 5.059137344360352,
+      "eval_negation-triplets_runtime": 1.0529,
+      "eval_negation-triplets_samples_per_second": 121.569,
+      "eval_negation-triplets_steps_per_second": 1.9,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_scitail-pairs-pos_loss": 1.9513375759124756,
+      "eval_scitail-pairs-pos_runtime": 1.2192,
+      "eval_scitail-pairs-pos_samples_per_second": 104.99,
+      "eval_scitail-pairs-pos_steps_per_second": 1.64,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_scitail-pairs-qa_loss": 2.7738723754882812,
+      "eval_scitail-pairs-qa_runtime": 0.7658,
+      "eval_scitail-pairs-qa_samples_per_second": 167.149,
+      "eval_scitail-pairs-qa_steps_per_second": 2.612,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_xsum-pairs_loss": 4.9543375968933105,
+      "eval_xsum-pairs_runtime": 3.9138,
+      "eval_xsum-pairs_samples_per_second": 32.705,
+      "eval_xsum-pairs_steps_per_second": 0.511,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_sciq_pairs_loss": 0.5856431722640991,
+      "eval_sciq_pairs_runtime": 5.846,
+      "eval_sciq_pairs_samples_per_second": 21.895,
+      "eval_sciq_pairs_steps_per_second": 0.342,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_qasc_pairs_loss": 4.644654750823975,
+      "eval_qasc_pairs_runtime": 0.9037,
+      "eval_qasc_pairs_samples_per_second": 141.642,
+      "eval_qasc_pairs_steps_per_second": 2.213,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_openbookqa_pairs_loss": 5.493581771850586,
+      "eval_openbookqa_pairs_runtime": 0.8617,
+      "eval_openbookqa_pairs_samples_per_second": 148.55,
+      "eval_openbookqa_pairs_steps_per_second": 2.321,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_msmarco_pairs_loss": 5.456345081329346,
+      "eval_msmarco_pairs_runtime": 1.9747,
+      "eval_msmarco_pairs_samples_per_second": 64.82,
+      "eval_msmarco_pairs_steps_per_second": 1.013,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_nq_pairs_loss": 5.471042633056641,
+      "eval_nq_pairs_runtime": 3.332,
+      "eval_nq_pairs_samples_per_second": 38.416,
+      "eval_nq_pairs_steps_per_second": 0.6,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_trivia_pairs_loss": 5.90379524230957,
+      "eval_trivia_pairs_runtime": 4.4746,
+      "eval_trivia_pairs_samples_per_second": 28.606,
+      "eval_trivia_pairs_steps_per_second": 0.447,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_gooaq_pairs_loss": 5.094766139984131,
+      "eval_gooaq_pairs_runtime": 1.4626,
+      "eval_gooaq_pairs_samples_per_second": 87.517,
+      "eval_gooaq_pairs_steps_per_second": 1.367,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_paws-pos_loss": 1.0914796590805054,
+      "eval_paws-pos_runtime": 1.0277,
+      "eval_paws-pos_samples_per_second": 124.555,
+      "eval_paws-pos_steps_per_second": 1.946,
+      "step": 160
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "eval_global_dataset_loss": 3.7029871940612793,
+      "eval_global_dataset_runtime": 19.3902,
+      "eval_global_dataset_samples_per_second": 21.454,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 160
+    },
+    {
+      "epoch": 0.12490302560124127,
+      "grad_norm": 12.19548511505127,
+      "learning_rate": 6.53936545240893e-06,
+      "loss": 4.0703,
+      "step": 161
+    },
+    {
+      "epoch": 0.1256788207913111,
+      "grad_norm": 11.81810188293457,
+      "learning_rate": 6.580493537015276e-06,
+      "loss": 3.7131,
+      "step": 162
+    },
+    {
+      "epoch": 0.1264546159813809,
+      "grad_norm": 12.847527503967285,
+      "learning_rate": 6.621621621621621e-06,
+      "loss": 3.8675,
+      "step": 163
+    },
+    {
+      "epoch": 0.12723041117145073,
+      "grad_norm": 12.332374572753906,
+      "learning_rate": 6.6627497062279665e-06,
+      "loss": 3.6557,
+      "step": 164
+    },
+    {
+      "epoch": 0.12800620636152055,
+      "grad_norm": 15.913554191589355,
+      "learning_rate": 6.703877790834312e-06,
+      "loss": 5.5114,
+      "step": 165
+    },
+    {
+      "epoch": 0.1287820015515904,
+      "grad_norm": 12.6423978805542,
+      "learning_rate": 6.745005875440658e-06,
+      "loss": 3.5924,
+      "step": 166
+    },
+    {
+      "epoch": 0.1295577967416602,
+      "grad_norm": 11.64156723022461,
+      "learning_rate": 6.786133960047002e-06,
+      "loss": 3.7331,
+      "step": 167
+    },
+    {
+      "epoch": 0.13033359193173003,
+      "grad_norm": 14.570131301879883,
+      "learning_rate": 6.8272620446533485e-06,
+      "loss": 5.2668,
+      "step": 168
+    },
+    {
+      "epoch": 0.13110938712179984,
+      "grad_norm": 13.375134468078613,
+      "learning_rate": 6.868390129259695e-06,
+      "loss": 3.5033,
+      "step": 169
+    },
+    {
+      "epoch": 0.13188518231186966,
+      "grad_norm": 13.718034744262695,
+      "learning_rate": 6.909518213866039e-06,
+      "loss": 3.9921,
+      "step": 170
+    },
+    {
+      "epoch": 0.13266097750193948,
+      "grad_norm": 12.68748950958252,
+      "learning_rate": 6.950646298472385e-06,
+      "loss": 3.3935,
+      "step": 171
+    },
+    {
+      "epoch": 0.13343677269200932,
+      "grad_norm": 18.421518325805664,
+      "learning_rate": 6.99177438307873e-06,
+      "loss": 4.9198,
+      "step": 172
+    },
+    {
+      "epoch": 0.13421256788207914,
+      "grad_norm": 21.028749465942383,
+      "learning_rate": 7.032902467685076e-06,
+      "loss": 1.3145,
+      "step": 173
+    },
+    {
+      "epoch": 0.13498836307214895,
+      "grad_norm": 21.025897979736328,
+      "learning_rate": 7.074030552291421e-06,
+      "loss": 5.0971,
+      "step": 174
+    },
+    {
+      "epoch": 0.13576415826221877,
+      "grad_norm": 16.524944305419922,
+      "learning_rate": 7.115158636897766e-06,
+      "loss": 4.7722,
+      "step": 175
+    },
+    {
+      "epoch": 0.1365399534522886,
+      "grad_norm": 15.925724029541016,
+      "learning_rate": 7.1562867215041125e-06,
+      "loss": 4.6407,
+      "step": 176
+    },
+    {
+      "epoch": 0.1373157486423584,
+      "grad_norm": 19.057390213012695,
+      "learning_rate": 7.197414806110458e-06,
+      "loss": 5.1164,
+      "step": 177
+    },
+    {
+      "epoch": 0.13809154383242825,
+      "grad_norm": 15.260638236999512,
+      "learning_rate": 7.238542890716803e-06,
+      "loss": 3.751,
+      "step": 178
+    },
+    {
+      "epoch": 0.13886733902249807,
+      "grad_norm": 30.054716110229492,
+      "learning_rate": 7.279670975323148e-06,
+      "loss": 5.4906,
+      "step": 179
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "grad_norm": 25.419301986694336,
+      "learning_rate": 7.3207990599294945e-06,
+      "loss": 5.1472,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_Qnli-dev_cosine_accuracy": 0.615234375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9118244647979736,
+      "eval_Qnli-dev_cosine_ap": 0.5757630053560099,
+      "eval_Qnli-dev_cosine_f1": 0.6347469220246238,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.708939254283905,
+      "eval_Qnli-dev_cosine_precision": 0.4686868686868687,
+      "eval_Qnli-dev_cosine_recall": 0.9830508474576272,
+      "eval_Qnli-dev_dot_accuracy": 0.57421875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 521.3780517578125,
+      "eval_Qnli-dev_dot_ap": 0.5092251488391427,
+      "eval_Qnli-dev_dot_f1": 0.6317204301075269,
+      "eval_Qnli-dev_dot_f1_threshold": 351.0001220703125,
+      "eval_Qnli-dev_dot_precision": 0.4625984251968504,
+      "eval_Qnli-dev_dot_recall": 0.9957627118644068,
+      "eval_Qnli-dev_euclidean_accuracy": 0.609375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 10.459676742553711,
+      "eval_Qnli-dev_euclidean_ap": 0.5817988420936584,
+      "eval_Qnli-dev_euclidean_f1": 0.6318607764390897,
+      "eval_Qnli-dev_euclidean_f1_threshold": 21.422916412353516,
+      "eval_Qnli-dev_euclidean_precision": 0.461839530332681,
+      "eval_Qnli-dev_euclidean_recall": 1.0,
+      "eval_Qnli-dev_manhattan_accuracy": 0.619140625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 212.72662353515625,
+      "eval_Qnli-dev_manhattan_ap": 0.5901977757397192,
+      "eval_Qnli-dev_manhattan_f1": 0.6324324324324324,
+      "eval_Qnli-dev_manhattan_f1_threshold": 349.28887939453125,
+      "eval_Qnli-dev_manhattan_precision": 0.4642857142857143,
+      "eval_Qnli-dev_manhattan_recall": 0.9915254237288136,
+      "eval_Qnli-dev_max_accuracy": 0.619140625,
+      "eval_Qnli-dev_max_accuracy_threshold": 521.3780517578125,
+      "eval_Qnli-dev_max_ap": 0.5901977757397192,
+      "eval_Qnli-dev_max_f1": 0.6347469220246238,
+      "eval_Qnli-dev_max_f1_threshold": 351.0001220703125,
+      "eval_Qnli-dev_max_precision": 0.4686868686868687,
+      "eval_Qnli-dev_max_recall": 1.0,
+      "eval_allNLI-dev_cosine_accuracy": 0.681640625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9692702293395996,
+      "eval_allNLI-dev_cosine_ap": 0.43588706986821285,
+      "eval_allNLI-dev_cosine_f1": 0.5211505922165821,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8656014800071716,
+      "eval_allNLI-dev_cosine_precision": 0.3684210526315789,
+      "eval_allNLI-dev_cosine_recall": 0.8901734104046243,
+      "eval_allNLI-dev_dot_accuracy": 0.66796875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 785.7974243164062,
+      "eval_allNLI-dev_dot_ap": 0.36610448753393265,
+      "eval_allNLI-dev_dot_f1": 0.5118343195266272,
+      "eval_allNLI-dev_dot_f1_threshold": 447.95782470703125,
+      "eval_allNLI-dev_dot_precision": 0.34393638170974156,
+      "eval_allNLI-dev_dot_recall": 1.0,
+      "eval_allNLI-dev_euclidean_accuracy": 0.677734375,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.728440284729004,
+      "eval_allNLI-dev_euclidean_ap": 0.4317993856992305,
+      "eval_allNLI-dev_euclidean_f1": 0.5214626391096979,
+      "eval_allNLI-dev_euclidean_f1_threshold": 15.216902732849121,
+      "eval_allNLI-dev_euclidean_precision": 0.35964912280701755,
+      "eval_allNLI-dev_euclidean_recall": 0.9479768786127167,
+      "eval_allNLI-dev_manhattan_accuracy": 0.673828125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 101.38760375976562,
+      "eval_allNLI-dev_manhattan_ap": 0.43314824920268913,
+      "eval_allNLI-dev_manhattan_f1": 0.5157421289355323,
+      "eval_allNLI-dev_manhattan_f1_threshold": 286.70037841796875,
+      "eval_allNLI-dev_manhattan_precision": 0.3481781376518219,
+      "eval_allNLI-dev_manhattan_recall": 0.9942196531791907,
+      "eval_allNLI-dev_max_accuracy": 0.681640625,
+      "eval_allNLI-dev_max_accuracy_threshold": 785.7974243164062,
+      "eval_allNLI-dev_max_ap": 0.43588706986821285,
+      "eval_allNLI-dev_max_f1": 0.5214626391096979,
+      "eval_allNLI-dev_max_f1_threshold": 447.95782470703125,
+      "eval_allNLI-dev_max_precision": 0.3684210526315789,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.5901977757397192,
+      "eval_sts-test_pearson_cosine": 0.20493152812453436,
+      "eval_sts-test_pearson_dot": 0.21775282373018834,
+      "eval_sts-test_pearson_euclidean": 0.2301348094949043,
+      "eval_sts-test_pearson_manhattan": 0.23840753470415393,
+      "eval_sts-test_pearson_max": 0.23840753470415393,
+      "eval_sts-test_spearman_cosine": 0.28828077894040277,
+      "eval_sts-test_spearman_dot": 0.23061179744080743,
+      "eval_sts-test_spearman_euclidean": 0.26600149771047865,
+      "eval_sts-test_spearman_manhattan": 0.26965490118794977,
+      "eval_sts-test_spearman_max": 0.28828077894040277,
+      "eval_vitaminc-pairs_loss": 3.7227895259857178,
+      "eval_vitaminc-pairs_runtime": 5.4557,
+      "eval_vitaminc-pairs_samples_per_second": 23.462,
+      "eval_vitaminc-pairs_steps_per_second": 0.367,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_negation-triplets_loss": 4.880922794342041,
+      "eval_negation-triplets_runtime": 1.0642,
+      "eval_negation-triplets_samples_per_second": 120.279,
+      "eval_negation-triplets_steps_per_second": 1.879,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_scitail-pairs-pos_loss": 1.6082611083984375,
+      "eval_scitail-pairs-pos_runtime": 1.2328,
+      "eval_scitail-pairs-pos_samples_per_second": 103.83,
+      "eval_scitail-pairs-pos_steps_per_second": 1.622,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_scitail-pairs-qa_loss": 2.3117823600769043,
+      "eval_scitail-pairs-qa_runtime": 0.763,
+      "eval_scitail-pairs-qa_samples_per_second": 167.752,
+      "eval_scitail-pairs-qa_steps_per_second": 2.621,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_xsum-pairs_loss": 4.128580093383789,
+      "eval_xsum-pairs_runtime": 3.9087,
+      "eval_xsum-pairs_samples_per_second": 32.748,
+      "eval_xsum-pairs_steps_per_second": 0.512,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_sciq_pairs_loss": 0.4913565516471863,
+      "eval_sciq_pairs_runtime": 5.8016,
+      "eval_sciq_pairs_samples_per_second": 22.063,
+      "eval_sciq_pairs_steps_per_second": 0.345,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_qasc_pairs_loss": 4.587019920349121,
+      "eval_qasc_pairs_runtime": 0.8933,
+      "eval_qasc_pairs_samples_per_second": 143.294,
+      "eval_qasc_pairs_steps_per_second": 2.239,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_openbookqa_pairs_loss": 5.624441146850586,
+      "eval_openbookqa_pairs_runtime": 0.852,
+      "eval_openbookqa_pairs_samples_per_second": 150.235,
+      "eval_openbookqa_pairs_steps_per_second": 2.347,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_msmarco_pairs_loss": 5.057087421417236,
+      "eval_msmarco_pairs_runtime": 1.965,
+      "eval_msmarco_pairs_samples_per_second": 65.14,
+      "eval_msmarco_pairs_steps_per_second": 1.018,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_nq_pairs_loss": 5.1600823402404785,
+      "eval_nq_pairs_runtime": 3.3352,
+      "eval_nq_pairs_samples_per_second": 38.378,
+      "eval_nq_pairs_steps_per_second": 0.6,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_trivia_pairs_loss": 5.498032093048096,
+      "eval_trivia_pairs_runtime": 4.4801,
+      "eval_trivia_pairs_samples_per_second": 28.57,
+      "eval_trivia_pairs_steps_per_second": 0.446,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_gooaq_pairs_loss": 4.9159464836120605,
+      "eval_gooaq_pairs_runtime": 1.4649,
+      "eval_gooaq_pairs_samples_per_second": 87.38,
+      "eval_gooaq_pairs_steps_per_second": 1.365,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_paws-pos_loss": 0.44706642627716064,
+      "eval_paws-pos_runtime": 1.0313,
+      "eval_paws-pos_samples_per_second": 124.114,
+      "eval_paws-pos_steps_per_second": 1.939,
+      "step": 180
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "eval_global_dataset_loss": 3.263042688369751,
+      "eval_global_dataset_runtime": 19.4013,
+      "eval_global_dataset_samples_per_second": 21.442,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 180
+    },
+    {
+      "epoch": 0.1404189294026377,
+      "grad_norm": 17.023818969726562,
+      "learning_rate": 7.361927144535839e-06,
+      "loss": 4.17,
+      "step": 181
+    },
+    {
+      "epoch": 0.14119472459270752,
+      "grad_norm": 22.472793579101562,
+      "learning_rate": 7.403055229142185e-06,
+      "loss": 4.8977,
+      "step": 182
+    },
+    {
+      "epoch": 0.14197051978277736,
+      "grad_norm": 14.055580139160156,
+      "learning_rate": 7.444183313748531e-06,
+      "loss": 3.5031,
+      "step": 183
+    },
+    {
+      "epoch": 0.14274631497284718,
+      "grad_norm": 14.841517448425293,
+      "learning_rate": 7.485311398354876e-06,
+      "loss": 3.4286,
+      "step": 184
+    },
+    {
+      "epoch": 0.143522110162917,
+      "grad_norm": 21.218738555908203,
+      "learning_rate": 7.526439482961222e-06,
+      "loss": 3.2536,
+      "step": 185
+    },
+    {
+      "epoch": 0.1442979053529868,
+      "grad_norm": 14.635804176330566,
+      "learning_rate": 7.567567567567567e-06,
+      "loss": 4.1632,
+      "step": 186
+    },
+    {
+      "epoch": 0.14507370054305663,
+      "grad_norm": 16.717111587524414,
+      "learning_rate": 7.608695652173912e-06,
+      "loss": 3.6512,
+      "step": 187
+    },
+    {
+      "epoch": 0.14584949573312644,
+      "grad_norm": 20.993993759155273,
+      "learning_rate": 7.649823736780258e-06,
+      "loss": 3.3453,
+      "step": 188
+    },
+    {
+      "epoch": 0.1466252909231963,
+      "grad_norm": 21.484180450439453,
+      "learning_rate": 7.690951821386603e-06,
+      "loss": 4.9785,
+      "step": 189
+    },
+    {
+      "epoch": 0.1474010861132661,
+      "grad_norm": 14.600364685058594,
+      "learning_rate": 7.732079905992948e-06,
+      "loss": 3.1781,
+      "step": 190
+    },
+    {
+      "epoch": 0.14817688130333592,
+      "grad_norm": 17.323484420776367,
+      "learning_rate": 7.773207990599295e-06,
+      "loss": 3.6681,
+      "step": 191
+    },
+    {
+      "epoch": 0.14895267649340574,
+      "grad_norm": 17.703405380249023,
+      "learning_rate": 7.814336075205639e-06,
+      "loss": 4.5109,
+      "step": 192
+    },
+    {
+      "epoch": 0.14972847168347556,
+      "grad_norm": 15.650117874145508,
+      "learning_rate": 7.855464159811986e-06,
+      "loss": 4.4412,
+      "step": 193
+    },
+    {
+      "epoch": 0.15050426687354537,
+      "grad_norm": 30.876625061035156,
+      "learning_rate": 7.896592244418331e-06,
+      "loss": 5.1287,
+      "step": 194
+    },
+    {
+      "epoch": 0.15128006206361522,
+      "grad_norm": 18.507810592651367,
+      "learning_rate": 7.937720329024676e-06,
+      "loss": 4.2127,
+      "step": 195
+    },
+    {
+      "epoch": 0.15205585725368503,
+      "grad_norm": 14.450047492980957,
+      "learning_rate": 7.978848413631022e-06,
+      "loss": 3.1766,
+      "step": 196
+    },
+    {
+      "epoch": 0.15283165244375485,
+      "grad_norm": 20.77067756652832,
+      "learning_rate": 8.019976498237367e-06,
+      "loss": 2.5901,
+      "step": 197
+    },
+    {
+      "epoch": 0.15360744763382467,
+      "grad_norm": 12.4572114944458,
+      "learning_rate": 8.061104582843712e-06,
+      "loss": 3.5886,
+      "step": 198
+    },
+    {
+      "epoch": 0.15438324282389448,
+      "grad_norm": 15.76103687286377,
+      "learning_rate": 8.102232667450058e-06,
+      "loss": 2.9085,
+      "step": 199
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "grad_norm": 14.595601081848145,
+      "learning_rate": 8.143360752056405e-06,
+      "loss": 2.4958,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_Qnli-dev_cosine_accuracy": 0.615234375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8780630826950073,
+      "eval_Qnli-dev_cosine_ap": 0.6112556879875849,
+      "eval_Qnli-dev_cosine_f1": 0.6363636363636364,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7175576090812683,
+      "eval_Qnli-dev_cosine_precision": 0.4714285714285714,
+      "eval_Qnli-dev_cosine_recall": 0.9788135593220338,
+      "eval_Qnli-dev_dot_accuracy": 0.580078125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 530.33447265625,
+      "eval_Qnli-dev_dot_ap": 0.5372984783278565,
+      "eval_Qnli-dev_dot_f1": 0.6305555555555555,
+      "eval_Qnli-dev_dot_f1_threshold": 403.0582275390625,
+      "eval_Qnli-dev_dot_precision": 0.4690082644628099,
+      "eval_Qnli-dev_dot_recall": 0.961864406779661,
+      "eval_Qnli-dev_euclidean_accuracy": 0.619140625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 12.641845703125,
+      "eval_Qnli-dev_euclidean_ap": 0.6133386902087727,
+      "eval_Qnli-dev_euclidean_f1": 0.6331521739130436,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.538082122802734,
+      "eval_Qnli-dev_euclidean_precision": 0.466,
+      "eval_Qnli-dev_euclidean_recall": 0.9872881355932204,
+      "eval_Qnli-dev_manhattan_accuracy": 0.619140625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 190.95291137695312,
+      "eval_Qnli-dev_manhattan_ap": 0.6200633454752746,
+      "eval_Qnli-dev_manhattan_f1": 0.6364883401920438,
+      "eval_Qnli-dev_manhattan_f1_threshold": 324.4531555175781,
+      "eval_Qnli-dev_manhattan_precision": 0.47058823529411764,
+      "eval_Qnli-dev_manhattan_recall": 0.9830508474576272,
+      "eval_Qnli-dev_max_accuracy": 0.619140625,
+      "eval_Qnli-dev_max_accuracy_threshold": 530.33447265625,
+      "eval_Qnli-dev_max_ap": 0.6200633454752746,
+      "eval_Qnli-dev_max_f1": 0.6364883401920438,
+      "eval_Qnli-dev_max_f1_threshold": 403.0582275390625,
+      "eval_Qnli-dev_max_precision": 0.4714285714285714,
+      "eval_Qnli-dev_max_recall": 0.9872881355932204,
+      "eval_allNLI-dev_cosine_accuracy": 0.6875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.976324200630188,
+      "eval_allNLI-dev_cosine_ap": 0.45960393014401535,
+      "eval_allNLI-dev_cosine_f1": 0.5253164556962024,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8385776877403259,
+      "eval_allNLI-dev_cosine_precision": 0.3616557734204793,
+      "eval_allNLI-dev_cosine_recall": 0.9595375722543352,
+      "eval_allNLI-dev_dot_accuracy": 0.677734375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 772.0188598632812,
+      "eval_allNLI-dev_dot_ap": 0.3804150461800332,
+      "eval_allNLI-dev_dot_f1": 0.5080763582966226,
+      "eval_allNLI-dev_dot_f1_threshold": 444.7850646972656,
+      "eval_allNLI-dev_dot_precision": 0.3405511811023622,
+      "eval_allNLI-dev_dot_recall": 1.0,
+      "eval_allNLI-dev_euclidean_accuracy": 0.685546875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.875195503234863,
+      "eval_allNLI-dev_euclidean_ap": 0.4551932697343415,
+      "eval_allNLI-dev_euclidean_f1": 0.5257903494176372,
+      "eval_allNLI-dev_euclidean_f1_threshold": 13.960739135742188,
+      "eval_allNLI-dev_euclidean_precision": 0.3691588785046729,
+      "eval_allNLI-dev_euclidean_recall": 0.9132947976878613,
+      "eval_allNLI-dev_manhattan_accuracy": 0.681640625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 96.69795227050781,
+      "eval_allNLI-dev_manhattan_ap": 0.4537023325460285,
+      "eval_allNLI-dev_manhattan_f1": 0.5204081632653061,
+      "eval_allNLI-dev_manhattan_f1_threshold": 222.21884155273438,
+      "eval_allNLI-dev_manhattan_precision": 0.3686746987951807,
+      "eval_allNLI-dev_manhattan_recall": 0.884393063583815,
+      "eval_allNLI-dev_max_accuracy": 0.6875,
+      "eval_allNLI-dev_max_accuracy_threshold": 772.0188598632812,
+      "eval_allNLI-dev_max_ap": 0.45960393014401535,
+      "eval_allNLI-dev_max_f1": 0.5257903494176372,
+      "eval_allNLI-dev_max_f1_threshold": 444.7850646972656,
+      "eval_allNLI-dev_max_precision": 0.3691588785046729,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6200633454752746,
+      "eval_sts-test_pearson_cosine": 0.24924277969472153,
+      "eval_sts-test_pearson_dot": 0.20632094538884468,
+      "eval_sts-test_pearson_euclidean": 0.2775041116953867,
+      "eval_sts-test_pearson_manhattan": 0.29073367102351505,
+      "eval_sts-test_pearson_max": 0.29073367102351505,
+      "eval_sts-test_spearman_cosine": 0.3391141842675074,
+      "eval_sts-test_spearman_dot": 0.21011509320588695,
+      "eval_sts-test_spearman_euclidean": 0.31565870563443316,
+      "eval_sts-test_spearman_manhattan": 0.3219084430440491,
+      "eval_sts-test_spearman_max": 0.3391141842675074,
+      "eval_vitaminc-pairs_loss": 3.716855764389038,
+      "eval_vitaminc-pairs_runtime": 5.4756,
+      "eval_vitaminc-pairs_samples_per_second": 23.377,
+      "eval_vitaminc-pairs_steps_per_second": 0.365,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_negation-triplets_loss": 4.766229152679443,
+      "eval_negation-triplets_runtime": 1.0705,
+      "eval_negation-triplets_samples_per_second": 119.57,
+      "eval_negation-triplets_steps_per_second": 1.868,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_scitail-pairs-pos_loss": 0.9834614992141724,
+      "eval_scitail-pairs-pos_runtime": 1.2447,
+      "eval_scitail-pairs-pos_samples_per_second": 102.839,
+      "eval_scitail-pairs-pos_steps_per_second": 1.607,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_scitail-pairs-qa_loss": 1.3808467388153076,
+      "eval_scitail-pairs-qa_runtime": 0.7767,
+      "eval_scitail-pairs-qa_samples_per_second": 164.794,
+      "eval_scitail-pairs-qa_steps_per_second": 2.575,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_xsum-pairs_loss": 3.506582260131836,
+      "eval_xsum-pairs_runtime": 3.9348,
+      "eval_xsum-pairs_samples_per_second": 32.53,
+      "eval_xsum-pairs_steps_per_second": 0.508,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_sciq_pairs_loss": 0.38454076647758484,
+      "eval_sciq_pairs_runtime": 5.8783,
+      "eval_sciq_pairs_samples_per_second": 21.775,
+      "eval_sciq_pairs_steps_per_second": 0.34,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_qasc_pairs_loss": 3.758909225463867,
+      "eval_qasc_pairs_runtime": 0.9063,
+      "eval_qasc_pairs_samples_per_second": 141.232,
+      "eval_qasc_pairs_steps_per_second": 2.207,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_openbookqa_pairs_loss": 4.948967456817627,
+      "eval_openbookqa_pairs_runtime": 0.8606,
+      "eval_openbookqa_pairs_samples_per_second": 148.737,
+      "eval_openbookqa_pairs_steps_per_second": 2.324,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_msmarco_pairs_loss": 4.785572528839111,
+      "eval_msmarco_pairs_runtime": 1.9606,
+      "eval_msmarco_pairs_samples_per_second": 65.286,
+      "eval_msmarco_pairs_steps_per_second": 1.02,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_nq_pairs_loss": 4.675125598907471,
+      "eval_nq_pairs_runtime": 3.3155,
+      "eval_nq_pairs_samples_per_second": 38.607,
+      "eval_nq_pairs_steps_per_second": 0.603,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_trivia_pairs_loss": 4.418967247009277,
+      "eval_trivia_pairs_runtime": 4.478,
+      "eval_trivia_pairs_samples_per_second": 28.584,
+      "eval_trivia_pairs_steps_per_second": 0.447,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_gooaq_pairs_loss": 4.486730575561523,
+      "eval_gooaq_pairs_runtime": 1.4616,
+      "eval_gooaq_pairs_samples_per_second": 87.578,
+      "eval_gooaq_pairs_steps_per_second": 1.368,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_paws-pos_loss": 0.4021705687046051,
+      "eval_paws-pos_runtime": 1.0276,
+      "eval_paws-pos_samples_per_second": 124.562,
+      "eval_paws-pos_steps_per_second": 1.946,
+      "step": 200
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_global_dataset_loss": 2.6410105228424072,
+      "eval_global_dataset_runtime": 19.3933,
+      "eval_global_dataset_samples_per_second": 21.451,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 200
+    },
+    {
+      "epoch": 0.15593483320403415,
+      "grad_norm": 36.21416091918945,
+      "learning_rate": 8.184488836662748e-06,
+      "loss": 5.5704,
+      "step": 201
+    },
+    {
+      "epoch": 0.15671062839410396,
+      "grad_norm": 15.138063430786133,
+      "learning_rate": 8.225616921269095e-06,
+      "loss": 2.8612,
+      "step": 202
+    },
+    {
+      "epoch": 0.15748642358417378,
+      "grad_norm": 27.27367401123047,
+      "learning_rate": 8.26674500587544e-06,
+      "loss": 4.8846,
+      "step": 203
+    },
+    {
+      "epoch": 0.1582622187742436,
+      "grad_norm": 19.24480438232422,
+      "learning_rate": 8.307873090481786e-06,
+      "loss": 3.1182,
+      "step": 204
+    },
+    {
+      "epoch": 0.1590380139643134,
+      "grad_norm": 18.584644317626953,
+      "learning_rate": 8.349001175088131e-06,
+      "loss": 1.9789,
+      "step": 205
+    },
+    {
+      "epoch": 0.15981380915438323,
+      "grad_norm": 15.61733341217041,
+      "learning_rate": 8.390129259694476e-06,
+      "loss": 2.9743,
+      "step": 206
+    },
+    {
+      "epoch": 0.16058960434445307,
+      "grad_norm": 16.221357345581055,
+      "learning_rate": 8.431257344300822e-06,
+      "loss": 3.6543,
+      "step": 207
+    },
+    {
+      "epoch": 0.1613653995345229,
+      "grad_norm": 13.963678359985352,
+      "learning_rate": 8.472385428907167e-06,
+      "loss": 3.117,
+      "step": 208
+    },
+    {
+      "epoch": 0.1621411947245927,
+      "grad_norm": 18.764089584350586,
+      "learning_rate": 8.513513513513514e-06,
+      "loss": 1.5977,
+      "step": 209
+    },
+    {
+      "epoch": 0.16291698991466252,
+      "grad_norm": 29.377960205078125,
+      "learning_rate": 8.554641598119857e-06,
+      "loss": 5.1443,
+      "step": 210
+    },
+    {
+      "epoch": 0.16369278510473234,
+      "grad_norm": 13.549287796020508,
+      "learning_rate": 8.595769682726204e-06,
+      "loss": 0.668,
+      "step": 211
+    },
+    {
+      "epoch": 0.16446858029480219,
+      "grad_norm": 14.000697135925293,
+      "learning_rate": 8.636897767332548e-06,
+      "loss": 0.7522,
+      "step": 212
+    },
+    {
+      "epoch": 0.165244375484872,
+      "grad_norm": 18.982425689697266,
+      "learning_rate": 8.678025851938895e-06,
+      "loss": 2.3447,
+      "step": 213
+    },
+    {
+      "epoch": 0.16602017067494182,
+      "grad_norm": 13.098822593688965,
+      "learning_rate": 8.71915393654524e-06,
+      "loss": 0.5818,
+      "step": 214
+    },
+    {
+      "epoch": 0.16679596586501164,
+      "grad_norm": 17.622783660888672,
+      "learning_rate": 8.760282021151586e-06,
+      "loss": 3.556,
+      "step": 215
+    },
+    {
+      "epoch": 0.16757176105508145,
+      "grad_norm": 24.308530807495117,
+      "learning_rate": 8.801410105757933e-06,
+      "loss": 2.8425,
+      "step": 216
+    },
+    {
+      "epoch": 0.16834755624515127,
+      "grad_norm": 41.18635559082031,
+      "learning_rate": 8.842538190364276e-06,
+      "loss": 5.4947,
+      "step": 217
+    },
+    {
+      "epoch": 0.16912335143522111,
+      "grad_norm": 38.11179733276367,
+      "learning_rate": 8.883666274970621e-06,
+      "loss": 4.2956,
+      "step": 218
+    },
+    {
+      "epoch": 0.16989914662529093,
+      "grad_norm": 20.423124313354492,
+      "learning_rate": 8.924794359576967e-06,
+      "loss": 2.1325,
+      "step": 219
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "grad_norm": 19.0192813873291,
+      "learning_rate": 8.965922444183314e-06,
+      "loss": 1.4286,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_Qnli-dev_cosine_accuracy": 0.625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8608779907226562,
+      "eval_Qnli-dev_cosine_ap": 0.6155996680907755,
+      "eval_Qnli-dev_cosine_f1": 0.6555183946488294,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7760155200958252,
+      "eval_Qnli-dev_cosine_precision": 0.5414364640883977,
+      "eval_Qnli-dev_cosine_recall": 0.8305084745762712,
+      "eval_Qnli-dev_dot_accuracy": 0.62109375,
+      "eval_Qnli-dev_dot_accuracy_threshold": 477.1441650390625,
+      "eval_Qnli-dev_dot_ap": 0.5823923648803983,
+      "eval_Qnli-dev_dot_f1": 0.6345029239766081,
+      "eval_Qnli-dev_dot_f1_threshold": 358.5615234375,
+      "eval_Qnli-dev_dot_precision": 0.484375,
+      "eval_Qnli-dev_dot_recall": 0.9194915254237288,
+      "eval_Qnli-dev_euclidean_accuracy": 0.625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 13.561055183410645,
+      "eval_Qnli-dev_euclidean_ap": 0.6128136787518578,
+      "eval_Qnli-dev_euclidean_f1": 0.6476510067114094,
+      "eval_Qnli-dev_euclidean_f1_threshold": 15.50450325012207,
+      "eval_Qnli-dev_euclidean_precision": 0.5361111111111111,
+      "eval_Qnli-dev_euclidean_recall": 0.8177966101694916,
+      "eval_Qnli-dev_manhattan_accuracy": 0.619140625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 215.1610565185547,
+      "eval_Qnli-dev_manhattan_ap": 0.621124218438047,
+      "eval_Qnli-dev_manhattan_f1": 0.645484949832776,
+      "eval_Qnli-dev_manhattan_f1_threshold": 273.8544921875,
+      "eval_Qnli-dev_manhattan_precision": 0.5331491712707183,
+      "eval_Qnli-dev_manhattan_recall": 0.8177966101694916,
+      "eval_Qnli-dev_max_accuracy": 0.625,
+      "eval_Qnli-dev_max_accuracy_threshold": 477.1441650390625,
+      "eval_Qnli-dev_max_ap": 0.621124218438047,
+      "eval_Qnli-dev_max_f1": 0.6555183946488294,
+      "eval_Qnli-dev_max_f1_threshold": 358.5615234375,
+      "eval_Qnli-dev_max_precision": 0.5414364640883977,
+      "eval_Qnli-dev_max_recall": 0.9194915254237288,
+      "eval_allNLI-dev_cosine_accuracy": 0.685546875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9713066816329956,
+      "eval_allNLI-dev_cosine_ap": 0.48223591780287633,
+      "eval_allNLI-dev_cosine_f1": 0.5408805031446541,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8952139616012573,
+      "eval_allNLI-dev_cosine_precision": 0.4243421052631579,
+      "eval_allNLI-dev_cosine_recall": 0.7456647398843931,
+      "eval_allNLI-dev_dot_accuracy": 0.677734375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 741.37939453125,
+      "eval_allNLI-dev_dot_ap": 0.3936517100476376,
+      "eval_allNLI-dev_dot_f1": 0.5074626865671642,
+      "eval_allNLI-dev_dot_f1_threshold": 416.87164306640625,
+      "eval_allNLI-dev_dot_precision": 0.3420523138832998,
+      "eval_allNLI-dev_dot_recall": 0.9826589595375722,
+      "eval_allNLI-dev_euclidean_accuracy": 0.685546875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.748725891113281,
+      "eval_allNLI-dev_euclidean_ap": 0.47719987976648454,
+      "eval_allNLI-dev_euclidean_f1": 0.5415860735009671,
+      "eval_allNLI-dev_euclidean_f1_threshold": 12.946854591369629,
+      "eval_allNLI-dev_euclidean_precision": 0.4069767441860465,
+      "eval_allNLI-dev_euclidean_recall": 0.8092485549132948,
+      "eval_allNLI-dev_manhattan_accuracy": 0.685546875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 110.15467834472656,
+      "eval_allNLI-dev_manhattan_ap": 0.4846184039440549,
+      "eval_allNLI-dev_manhattan_f1": 0.5335753176043557,
+      "eval_allNLI-dev_manhattan_f1_threshold": 234.3075408935547,
+      "eval_allNLI-dev_manhattan_precision": 0.3888888888888889,
+      "eval_allNLI-dev_manhattan_recall": 0.8497109826589595,
+      "eval_allNLI-dev_max_accuracy": 0.685546875,
+      "eval_allNLI-dev_max_accuracy_threshold": 741.37939453125,
+      "eval_allNLI-dev_max_ap": 0.4846184039440549,
+      "eval_allNLI-dev_max_f1": 0.5415860735009671,
+      "eval_allNLI-dev_max_f1_threshold": 416.87164306640625,
+      "eval_allNLI-dev_max_precision": 0.4243421052631579,
+      "eval_allNLI-dev_max_recall": 0.9826589595375722,
+      "eval_sequential_score": 0.621124218438047,
+      "eval_sts-test_pearson_cosine": 0.31919410971368356,
+      "eval_sts-test_pearson_dot": 0.21556048485576235,
+      "eval_sts-test_pearson_euclidean": 0.35723202726898373,
+      "eval_sts-test_pearson_manhattan": 0.36765421659339287,
+      "eval_sts-test_pearson_max": 0.36765421659339287,
+      "eval_sts-test_spearman_cosine": 0.41124188237768894,
+      "eval_sts-test_spearman_dot": 0.2248200891915325,
+      "eval_sts-test_spearman_euclidean": 0.389006828577642,
+      "eval_sts-test_spearman_manhattan": 0.39845426736998035,
+      "eval_sts-test_spearman_max": 0.41124188237768894,
+      "eval_vitaminc-pairs_loss": 4.022590160369873,
+      "eval_vitaminc-pairs_runtime": 5.4704,
+      "eval_vitaminc-pairs_samples_per_second": 23.399,
+      "eval_vitaminc-pairs_steps_per_second": 0.366,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_negation-triplets_loss": 4.421084403991699,
+      "eval_negation-triplets_runtime": 1.0619,
+      "eval_negation-triplets_samples_per_second": 120.534,
+      "eval_negation-triplets_steps_per_second": 1.883,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_scitail-pairs-pos_loss": 0.7828177809715271,
+      "eval_scitail-pairs-pos_runtime": 1.2253,
+      "eval_scitail-pairs-pos_samples_per_second": 104.468,
+      "eval_scitail-pairs-pos_steps_per_second": 1.632,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_scitail-pairs-qa_loss": 0.9572672843933105,
+      "eval_scitail-pairs-qa_runtime": 0.7677,
+      "eval_scitail-pairs-qa_samples_per_second": 166.739,
+      "eval_scitail-pairs-qa_steps_per_second": 2.605,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_xsum-pairs_loss": 3.0789806842803955,
+      "eval_xsum-pairs_runtime": 3.9039,
+      "eval_xsum-pairs_samples_per_second": 32.788,
+      "eval_xsum-pairs_steps_per_second": 0.512,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_sciq_pairs_loss": 0.39149752259254456,
+      "eval_sciq_pairs_runtime": 5.8117,
+      "eval_sciq_pairs_samples_per_second": 22.024,
+      "eval_sciq_pairs_steps_per_second": 0.344,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_qasc_pairs_loss": 3.374030828475952,
+      "eval_qasc_pairs_runtime": 0.9033,
+      "eval_qasc_pairs_samples_per_second": 141.696,
+      "eval_qasc_pairs_steps_per_second": 2.214,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_openbookqa_pairs_loss": 5.161572456359863,
+      "eval_openbookqa_pairs_runtime": 0.8619,
+      "eval_openbookqa_pairs_samples_per_second": 148.506,
+      "eval_openbookqa_pairs_steps_per_second": 2.32,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_msmarco_pairs_loss": 4.489230632781982,
+      "eval_msmarco_pairs_runtime": 1.9668,
+      "eval_msmarco_pairs_samples_per_second": 65.079,
+      "eval_msmarco_pairs_steps_per_second": 1.017,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_nq_pairs_loss": 4.549116611480713,
+      "eval_nq_pairs_runtime": 3.3346,
+      "eval_nq_pairs_samples_per_second": 38.385,
+      "eval_nq_pairs_steps_per_second": 0.6,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_trivia_pairs_loss": 5.130815029144287,
+      "eval_trivia_pairs_runtime": 4.4818,
+      "eval_trivia_pairs_samples_per_second": 28.56,
+      "eval_trivia_pairs_steps_per_second": 0.446,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_gooaq_pairs_loss": 3.8106689453125,
+      "eval_gooaq_pairs_runtime": 1.4609,
+      "eval_gooaq_pairs_samples_per_second": 87.615,
+      "eval_gooaq_pairs_steps_per_second": 1.369,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_paws-pos_loss": 0.1346072554588318,
+      "eval_paws-pos_runtime": 1.0308,
+      "eval_paws-pos_samples_per_second": 124.178,
+      "eval_paws-pos_steps_per_second": 1.94,
+      "step": 220
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "eval_global_dataset_loss": 2.3529915809631348,
+      "eval_global_dataset_runtime": 19.4389,
+      "eval_global_dataset_samples_per_second": 21.4,
+      "eval_global_dataset_steps_per_second": 0.36,
+      "step": 220
+    },
+    {
+      "epoch": 0.17145073700543056,
+      "grad_norm": 24.33696174621582,
+      "learning_rate": 9.007050528789659e-06,
+      "loss": 2.2834,
+      "step": 221
+    },
+    {
+      "epoch": 0.17222653219550038,
+      "grad_norm": 19.657947540283203,
+      "learning_rate": 9.048178613396003e-06,
+      "loss": 3.4539,
+      "step": 222
+    },
+    {
+      "epoch": 0.1730023273855702,
+      "grad_norm": 22.255239486694336,
+      "learning_rate": 9.08930669800235e-06,
+      "loss": 2.7994,
+      "step": 223
+    },
+    {
+      "epoch": 0.17377812257564004,
+      "grad_norm": 20.619884490966797,
+      "learning_rate": 9.130434782608695e-06,
+      "loss": 2.463,
+      "step": 224
+    },
+    {
+      "epoch": 0.17455391776570986,
+      "grad_norm": 22.244462966918945,
+      "learning_rate": 9.17156286721504e-06,
+      "loss": 4.1749,
+      "step": 225
+    },
+    {
+      "epoch": 0.17532971295577968,
+      "grad_norm": 25.1329402923584,
+      "learning_rate": 9.212690951821385e-06,
+      "loss": 3.7001,
+      "step": 226
+    },
+    {
+      "epoch": 0.1761055081458495,
+      "grad_norm": 35.8182258605957,
+      "learning_rate": 9.253819036427732e-06,
+      "loss": 5.6003,
+      "step": 227
+    },
+    {
+      "epoch": 0.1768813033359193,
+      "grad_norm": 33.197845458984375,
+      "learning_rate": 9.294947121034076e-06,
+      "loss": 5.5236,
+      "step": 228
+    },
+    {
+      "epoch": 0.17765709852598913,
+      "grad_norm": 14.53373908996582,
+      "learning_rate": 9.336075205640421e-06,
+      "loss": 2.8411,
+      "step": 229
+    },
+    {
+      "epoch": 0.17843289371605897,
+      "grad_norm": 8.925052642822266,
+      "learning_rate": 9.377203290246768e-06,
+      "loss": 0.3611,
+      "step": 230
+    },
+    {
+      "epoch": 0.1792086889061288,
+      "grad_norm": 19.080123901367188,
+      "learning_rate": 9.418331374853114e-06,
+      "loss": 3.4145,
+      "step": 231
+    },
+    {
+      "epoch": 0.1799844840961986,
+      "grad_norm": 16.67955780029297,
+      "learning_rate": 9.459459459459459e-06,
+      "loss": 2.7527,
+      "step": 232
+    },
+    {
+      "epoch": 0.18076027928626842,
+      "grad_norm": 13.526026725769043,
+      "learning_rate": 9.500587544065804e-06,
+      "loss": 1.977,
+      "step": 233
+    },
+    {
+      "epoch": 0.18153607447633824,
+      "grad_norm": 12.823989868164062,
+      "learning_rate": 9.54171562867215e-06,
+      "loss": 1.9694,
+      "step": 234
+    },
+    {
+      "epoch": 0.18231186966640806,
+      "grad_norm": 14.984912872314453,
+      "learning_rate": 9.582843713278495e-06,
+      "loss": 2.4804,
+      "step": 235
+    },
+    {
+      "epoch": 0.1830876648564779,
+      "grad_norm": 13.724047660827637,
+      "learning_rate": 9.62397179788484e-06,
+      "loss": 2.7869,
+      "step": 236
+    },
+    {
+      "epoch": 0.18386346004654772,
+      "grad_norm": 14.626879692077637,
+      "learning_rate": 9.665099882491187e-06,
+      "loss": 2.0562,
+      "step": 237
+    },
+    {
+      "epoch": 0.18463925523661753,
+      "grad_norm": 8.06078815460205,
+      "learning_rate": 9.706227967097532e-06,
+      "loss": 0.3609,
+      "step": 238
+    },
+    {
+      "epoch": 0.18541505042668735,
+      "grad_norm": 13.897974014282227,
+      "learning_rate": 9.747356051703878e-06,
+      "loss": 1.9205,
+      "step": 239
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "grad_norm": 6.743955612182617,
+      "learning_rate": 9.788484136310221e-06,
+      "loss": 0.2964,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_Qnli-dev_cosine_accuracy": 0.62890625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8693021535873413,
+      "eval_Qnli-dev_cosine_ap": 0.611781004288293,
+      "eval_Qnli-dev_cosine_f1": 0.6366559485530547,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7878068685531616,
+      "eval_Qnli-dev_cosine_precision": 0.5129533678756477,
+      "eval_Qnli-dev_cosine_recall": 0.8389830508474576,
+      "eval_Qnli-dev_dot_accuracy": 0.595703125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 521.8904418945312,
+      "eval_Qnli-dev_dot_ap": 0.5496339935508336,
+      "eval_Qnli-dev_dot_f1": 0.6397694524495677,
+      "eval_Qnli-dev_dot_f1_threshold": 375.82940673828125,
+      "eval_Qnli-dev_dot_precision": 0.4847161572052402,
+      "eval_Qnli-dev_dot_recall": 0.940677966101695,
+      "eval_Qnli-dev_euclidean_accuracy": 0.626953125,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 11.786832809448242,
+      "eval_Qnli-dev_euclidean_ap": 0.616692733684346,
+      "eval_Qnli-dev_euclidean_f1": 0.6368159203980099,
+      "eval_Qnli-dev_euclidean_f1_threshold": 15.11730670928955,
+      "eval_Qnli-dev_euclidean_precision": 0.5231607629427792,
+      "eval_Qnli-dev_euclidean_recall": 0.8135593220338984,
+      "eval_Qnli-dev_manhattan_accuracy": 0.6171875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 198.48147583007812,
+      "eval_Qnli-dev_manhattan_ap": 0.6210006336689851,
+      "eval_Qnli-dev_manhattan_f1": 0.6374367622259697,
+      "eval_Qnli-dev_manhattan_f1_threshold": 262.7891845703125,
+      "eval_Qnli-dev_manhattan_precision": 0.5294117647058824,
+      "eval_Qnli-dev_manhattan_recall": 0.8008474576271186,
+      "eval_Qnli-dev_max_accuracy": 0.62890625,
+      "eval_Qnli-dev_max_accuracy_threshold": 521.8904418945312,
+      "eval_Qnli-dev_max_ap": 0.6210006336689851,
+      "eval_Qnli-dev_max_f1": 0.6397694524495677,
+      "eval_Qnli-dev_max_f1_threshold": 375.82940673828125,
+      "eval_Qnli-dev_max_precision": 0.5294117647058824,
+      "eval_Qnli-dev_max_recall": 0.940677966101695,
+      "eval_allNLI-dev_cosine_accuracy": 0.6875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9714217782020569,
+      "eval_allNLI-dev_cosine_ap": 0.5054463009748529,
+      "eval_allNLI-dev_cosine_f1": 0.5534591194968553,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.891494631767273,
+      "eval_allNLI-dev_cosine_precision": 0.4342105263157895,
+      "eval_allNLI-dev_cosine_recall": 0.7630057803468208,
+      "eval_allNLI-dev_dot_accuracy": 0.6796875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 697.1558837890625,
+      "eval_allNLI-dev_dot_ap": 0.40155312352801564,
+      "eval_allNLI-dev_dot_f1": 0.5088235294117648,
+      "eval_allNLI-dev_dot_f1_threshold": 376.5243225097656,
+      "eval_allNLI-dev_dot_precision": 0.34122287968441817,
+      "eval_allNLI-dev_dot_recall": 1.0,
+      "eval_allNLI-dev_euclidean_accuracy": 0.685546875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.3458075523376465,
+      "eval_allNLI-dev_euclidean_ap": 0.5028194115939991,
+      "eval_allNLI-dev_euclidean_f1": 0.5518590998043053,
+      "eval_allNLI-dev_euclidean_f1_threshold": 12.797416687011719,
+      "eval_allNLI-dev_euclidean_precision": 0.4171597633136095,
+      "eval_allNLI-dev_euclidean_recall": 0.815028901734104,
+      "eval_allNLI-dev_manhattan_accuracy": 0.689453125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 145.72430419921875,
+      "eval_allNLI-dev_manhattan_ap": 0.5074540422577104,
+      "eval_allNLI-dev_manhattan_f1": 0.5382932166301969,
+      "eval_allNLI-dev_manhattan_f1_threshold": 210.015625,
+      "eval_allNLI-dev_manhattan_precision": 0.43309859154929575,
+      "eval_allNLI-dev_manhattan_recall": 0.7109826589595376,
+      "eval_allNLI-dev_max_accuracy": 0.689453125,
+      "eval_allNLI-dev_max_accuracy_threshold": 697.1558837890625,
+      "eval_allNLI-dev_max_ap": 0.5074540422577104,
+      "eval_allNLI-dev_max_f1": 0.5534591194968553,
+      "eval_allNLI-dev_max_f1_threshold": 376.5243225097656,
+      "eval_allNLI-dev_max_precision": 0.4342105263157895,
+      "eval_allNLI-dev_max_recall": 1.0,
+      "eval_sequential_score": 0.6210006336689851,
+      "eval_sts-test_pearson_cosine": 0.437261385872042,
+      "eval_sts-test_pearson_dot": 0.27944802216402886,
+      "eval_sts-test_pearson_euclidean": 0.46065093061659046,
+      "eval_sts-test_pearson_manhattan": 0.4698077558920942,
+      "eval_sts-test_pearson_max": 0.4698077558920942,
+      "eval_sts-test_spearman_cosine": 0.5068103175719304,
+      "eval_sts-test_spearman_dot": 0.28160415924837434,
+      "eval_sts-test_spearman_euclidean": 0.4813224156515044,
+      "eval_sts-test_spearman_manhattan": 0.4888016492165877,
+      "eval_sts-test_spearman_max": 0.5068103175719304,
+      "eval_vitaminc-pairs_loss": 3.8961689472198486,
+      "eval_vitaminc-pairs_runtime": 5.4898,
+      "eval_vitaminc-pairs_samples_per_second": 23.316,
+      "eval_vitaminc-pairs_steps_per_second": 0.364,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_negation-triplets_loss": 3.9956982135772705,
+      "eval_negation-triplets_runtime": 1.0718,
+      "eval_negation-triplets_samples_per_second": 119.43,
+      "eval_negation-triplets_steps_per_second": 1.866,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_scitail-pairs-pos_loss": 0.6250349283218384,
+      "eval_scitail-pairs-pos_runtime": 1.2283,
+      "eval_scitail-pairs-pos_samples_per_second": 104.211,
+      "eval_scitail-pairs-pos_steps_per_second": 1.628,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_scitail-pairs-qa_loss": 0.8983888030052185,
+      "eval_scitail-pairs-qa_runtime": 0.7728,
+      "eval_scitail-pairs-qa_samples_per_second": 165.636,
+      "eval_scitail-pairs-qa_steps_per_second": 2.588,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_xsum-pairs_loss": 2.681138038635254,
+      "eval_xsum-pairs_runtime": 3.921,
+      "eval_xsum-pairs_samples_per_second": 32.645,
+      "eval_xsum-pairs_steps_per_second": 0.51,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_sciq_pairs_loss": 0.3189742863178253,
+      "eval_sciq_pairs_runtime": 5.8916,
+      "eval_sciq_pairs_samples_per_second": 21.726,
+      "eval_sciq_pairs_steps_per_second": 0.339,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_qasc_pairs_loss": 2.76663875579834,
+      "eval_qasc_pairs_runtime": 0.9119,
+      "eval_qasc_pairs_samples_per_second": 140.374,
+      "eval_qasc_pairs_steps_per_second": 2.193,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_openbookqa_pairs_loss": 4.003782749176025,
+      "eval_openbookqa_pairs_runtime": 0.8645,
+      "eval_openbookqa_pairs_samples_per_second": 148.058,
+      "eval_openbookqa_pairs_steps_per_second": 2.313,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_msmarco_pairs_loss": 3.789357900619507,
+      "eval_msmarco_pairs_runtime": 1.9837,
+      "eval_msmarco_pairs_samples_per_second": 64.525,
+      "eval_msmarco_pairs_steps_per_second": 1.008,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_nq_pairs_loss": 3.925625801086426,
+      "eval_nq_pairs_runtime": 3.3218,
+      "eval_nq_pairs_samples_per_second": 38.534,
+      "eval_nq_pairs_steps_per_second": 0.602,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_trivia_pairs_loss": 3.9500904083251953,
+      "eval_trivia_pairs_runtime": 4.4861,
+      "eval_trivia_pairs_samples_per_second": 28.533,
+      "eval_trivia_pairs_steps_per_second": 0.446,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_gooaq_pairs_loss": 3.2876720428466797,
+      "eval_gooaq_pairs_runtime": 1.4821,
+      "eval_gooaq_pairs_samples_per_second": 86.362,
+      "eval_gooaq_pairs_steps_per_second": 1.349,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_paws-pos_loss": 0.12235681712627411,
+      "eval_paws-pos_runtime": 1.0363,
+      "eval_paws-pos_samples_per_second": 123.516,
+      "eval_paws-pos_steps_per_second": 1.93,
+      "step": 240
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "eval_global_dataset_loss": 1.9579764604568481,
+      "eval_global_dataset_runtime": 19.4,
+      "eval_global_dataset_samples_per_second": 21.443,
+      "eval_global_dataset_steps_per_second": 0.361,
+      "step": 240
+    },
+    {
+      "epoch": 0.186966640806827,
+      "grad_norm": 17.613515853881836,
+      "learning_rate": 9.829612220916568e-06,
+      "loss": 3.2402,
+      "step": 241
+    },
+    {
+      "epoch": 0.18774243599689683,
+      "grad_norm": 19.704069137573242,
+      "learning_rate": 9.870740305522913e-06,
+      "loss": 3.1076,
+      "step": 242
+    },
+    {
+      "epoch": 0.18851823118696664,
+      "grad_norm": 14.683489799499512,
+      "learning_rate": 9.911868390129259e-06,
+      "loss": 1.8656,
+      "step": 243
+    },
+    {
+      "epoch": 0.18929402637703646,
+      "grad_norm": 15.879508018493652,
+      "learning_rate": 9.952996474735606e-06,
+      "loss": 2.7912,
+      "step": 244
+    },
+    {
+      "epoch": 0.19006982156710628,
+      "grad_norm": 5.585937023162842,
+      "learning_rate": 9.99412455934195e-06,
+      "loss": 0.2569,
+      "step": 245
+    },
+    {
+      "epoch": 0.1908456167571761,
+      "grad_norm": 18.365917205810547,
+      "learning_rate": 1.0035252643948295e-05,
+      "loss": 2.1095,
+      "step": 246
+    },
+    {
+      "epoch": 0.19162141194724594,
+      "grad_norm": 14.385225296020508,
+      "learning_rate": 1.007638072855464e-05,
+      "loss": 1.5261,
+      "step": 247
+    },
+    {
+      "epoch": 0.19239720713731576,
+      "grad_norm": 23.97905921936035,
+      "learning_rate": 1.0117508813160987e-05,
+      "loss": 3.4551,
+      "step": 248
+    },
+    {
+      "epoch": 0.19317300232738557,
+      "grad_norm": 15.02582836151123,
+      "learning_rate": 1.0158636897767332e-05,
+      "loss": 2.2465,
+      "step": 249
+    },
+    {
+      "epoch": 0.1939487975174554,
+      "grad_norm": 32.1656494140625,
+      "learning_rate": 1.0199764982373677e-05,
+      "loss": 4.7511,
+      "step": 250
+    },
+    {
+      "epoch": 0.1947245927075252,
+      "grad_norm": 22.766891479492188,
+      "learning_rate": 1.0240893066980021e-05,
+      "loss": 2.9648,
+      "step": 251
+    },
+    {
+      "epoch": 0.19550038789759502,
+      "grad_norm": 22.01688575744629,
+      "learning_rate": 1.0282021151586368e-05,
+      "loss": 3.4702,
+      "step": 252
+    },
+    {
+      "epoch": 0.19627618308766487,
+      "grad_norm": 19.137022018432617,
+      "learning_rate": 1.0323149236192713e-05,
+      "loss": 2.2448,
+      "step": 253
+    },
+    {
+      "epoch": 0.19705197827773469,
+      "grad_norm": 26.455888748168945,
+      "learning_rate": 1.0364277320799059e-05,
+      "loss": 4.5872,
+      "step": 254
+    },
+    {
+      "epoch": 0.1978277734678045,
+      "grad_norm": 15.777259826660156,
+      "learning_rate": 1.0405405405405406e-05,
+      "loss": 1.8122,
+      "step": 255
+    },
+    {
+      "epoch": 0.19860356865787432,
+      "grad_norm": 17.67413902282715,
+      "learning_rate": 1.0446533490011751e-05,
+      "loss": 2.8551,
+      "step": 256
+    },
+    {
+      "epoch": 0.19937936384794414,
+      "grad_norm": 17.045608520507812,
+      "learning_rate": 1.0487661574618094e-05,
+      "loss": 2.6086,
+      "step": 257
+    },
+    {
+      "epoch": 0.20015515903801395,
+      "grad_norm": 14.350496292114258,
+      "learning_rate": 1.052878965922444e-05,
+      "loss": 1.232,
+      "step": 258
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2578,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 258,
+  "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": 96,
+  "trial_name": null,
+  "trial_params": null
+}