first commit

all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 2.0,
+    "eval_e": 0.0,
+    "eval_f1": 0.003218218203323381,
+    "eval_loss": 7.38798189163208,
+    "eval_runtime": 81.0639,
+    "eval_samples": 3576,
+    "eval_samples_per_second": 44.113,
+    "eval_steps_per_second": 44.113,
+    "test_e": 0.6023489932885906,
+    "test_f1": 0.5506494789202888,
+    "test_loss": 2.7486965656280518,
+    "test_runtime": 80.778,
+    "test_samples_per_second": 44.269,
+    "test_steps_per_second": 44.269,
+    "train_loss": 1.5977188166040575,
+    "train_runtime": 2455.1123,
+    "train_samples": 10138,
+    "train_samples_per_second": 8.259,
+    "train_steps_per_second": 8.259
+}

callback-metrics-state-0.json

@@ -0,0 +1 @@
+{"INDIRECT QUOTE_loss": 7.3920416831970215, "INDIRECT QUOTE_f1": 0.002867467424910844, "INDIRECT QUOTE_e": 0.0, "DIRECT QUOTE_loss": 7.395792007446289, "DIRECT QUOTE_f1": 0.0026685346041993878, "DIRECT QUOTE_e": 0.0, "BACKGROUND_loss": 7.365277290344238, "BACKGROUND_f1": 0.004162366856499275, "BACKGROUND_e": 0.0, "PUBLISHED WORK_loss": 7.38569974899292, "PUBLISHED WORK_f1": 0.0026257877273695965, "PUBLISHED WORK_e": 0.0, "STATEMENT_loss": 7.371387004852295, "STATEMENT_f1": 0.002811513597655576, "STATEMENT_e": 0.0, "SOCIAL MEDIA POST_loss": 7.371214389801025, "SOCIAL MEDIA POST_f1": 0.0021667480985203876, "SOCIAL MEDIA POST_e": 0.0, "PRESS REPORT_loss": 7.400311470031738, "PRESS REPORT_f1": 0.002322142960838986, "PRESS REPORT_e": 0.0, "DECLINED COMMENT_loss": 7.410003662109375, "DECLINED COMMENT_f1": 0.0035249913168271703, "DECLINED COMMENT_e": 0.0, "PROPOSAL/ORDER/LAW_loss": 7.406617641448975, "PROPOSAL/ORDER/LAW_f1": 0.008154212035879524, "PROPOSAL/ORDER/LAW_e": 0.0, "PRICE SIGNAL_loss": 7.322811126708984, "PRICE SIGNAL_f1": 0.0020276297755392704, "PRICE SIGNAL_e": 0.0, "NARRATIVE_loss": 7.39856481552124, "NARRATIVE_f1": 0.005181297276397006, "NARRATIVE_e": 0.0, "DIRECT OBSERVATION_loss": 7.415427207946777, "DIRECT OBSERVATION_f1": 0.0, "DIRECT OBSERVATION_e": 0.0, "COMMUNICATION_loss": 7.398114204406738, "COMMUNICATION_f1": 0.003210920804232077, "COMMUNICATION_e": 0.0, "PUBLIC SPEECH_loss": 7.275794506072998, "PUBLIC SPEECH_f1": 0.010951878692647335, "PUBLIC SPEECH_e": 0.0, "VOTE/POLL_loss": 7.344000816345215, "VOTE/POLL_f1": 0.0036740648664935344, "VOTE/POLL_e": 0.0, "COURT PROCEEDING_loss": 7.464186191558838, "COURT PROCEEDING_f1": 0.0020636257293816943, "COURT PROCEEDING_e": 0.0, "full_loss": 7.38798189163208, "full_f1": 0.003218218203323381, "full_e": 0.0}

callback-metrics-state-10000.json

@@ -0,0 +1 @@
+{"INDIRECT QUOTE_loss": 2.0357444286346436, "INDIRECT QUOTE_f1": 0.5565934558159891, "INDIRECT QUOTE_e": 0.5645454545454546, "DIRECT QUOTE_loss": 2.3745360374450684, "DIRECT QUOTE_f1": 0.5075596828055501, "DIRECT QUOTE_e": 0.5552784704904405, "BACKGROUND_loss": 2.600733995437622, "BACKGROUND_f1": 0.5894816406479484, "BACKGROUND_e": 0.6398104265402843, "PUBLISHED WORK_loss": 3.1265647411346436, "PUBLISHED WORK_f1": 0.40655165655496034, "PUBLISHED WORK_e": 0.44274809160305345, "STATEMENT_loss": 2.103498935699463, "STATEMENT_f1": 0.5028104423094844, "STATEMENT_e": 0.6381578947368421, "SOCIAL MEDIA POST_loss": 4.426347732543945, "SOCIAL MEDIA POST_f1": 0.1238095238095238, "SOCIAL MEDIA POST_e": 0.24242424242424243, "PRESS REPORT_loss": 3.8396220207214355, "PRESS REPORT_f1": 0.4449627001351139, "PRESS REPORT_e": 0.4827586206896552, "DECLINED COMMENT_loss": 2.6896440982818604, "DECLINED COMMENT_f1": 0.3125, "DECLINED COMMENT_e": 0.4375, "PROPOSAL/ORDER/LAW_loss": 3.2415931224823, "PROPOSAL/ORDER/LAW_f1": 0.28395954022988507, "PROPOSAL/ORDER/LAW_e": 0.36, "PRICE SIGNAL_loss": 3.032341241836548, "PRICE SIGNAL_f1": 0.39546112988894255, "PRICE SIGNAL_e": 0.42105263157894735, "NARRATIVE_loss": 1.632098913192749, "NARRATIVE_f1": 0.5818456841015806, "NARRATIVE_e": 0.6764705882352942, "DIRECT OBSERVATION_loss": 4.492552757263184, "DIRECT OBSERVATION_f1": 0.01694915254237288, "DIRECT OBSERVATION_e": 0.11864406779661017, "COMMUNICATION_loss": 2.633908748626709, "COMMUNICATION_f1": 0.546875, "COMMUNICATION_e": 0.546875, "PUBLIC SPEECH_loss": 3.456928014755249, "PUBLIC SPEECH_f1": 0.27166666666666667, "PUBLIC SPEECH_e": 0.6333333333333333, "VOTE/POLL_loss": 4.232885360717773, "VOTE/POLL_f1": 0.29275160256595884, "VOTE/POLL_e": 0.30434782608695654, "COURT PROCEEDING_loss": 3.8775787353515625, "COURT PROCEEDING_f1": 0.3294430014430014, "COURT PROCEEDING_e": 0.44, "full_loss": 2.430802822113037, "full_f1": 0.5065431932934646, "full_e": 0.5542505592841164}

callback-metrics-state-15000.json

@@ -0,0 +1 @@
+{"INDIRECT QUOTE_loss": 2.478729248046875, "INDIRECT QUOTE_f1": 0.5864855228766466, "INDIRECT QUOTE_e": 0.5954545454545455, "DIRECT QUOTE_loss": 3.0114314556121826, "DIRECT QUOTE_f1": 0.530530760516083, "DIRECT QUOTE_e": 0.5685785536159601, "BACKGROUND_loss": 3.5276763439178467, "BACKGROUND_f1": 0.5905275636316465, "BACKGROUND_e": 0.6350710900473934, "PUBLISHED WORK_loss": 4.573206901550293, "PUBLISHED WORK_f1": 0.427382933012704, "PUBLISHED WORK_e": 0.48854961832061067, "STATEMENT_loss": 2.3900763988494873, "STATEMENT_f1": 0.5186403508771931, "STATEMENT_e": 0.6447368421052632, "SOCIAL MEDIA POST_loss": 5.451953887939453, "SOCIAL MEDIA POST_f1": 0.18181818181818182, "SOCIAL MEDIA POST_e": 0.30303030303030304, "PRESS REPORT_loss": 5.020430564880371, "PRESS REPORT_f1": 0.5002736726874658, "PRESS REPORT_e": 0.5344827586206896, "DECLINED COMMENT_loss": 3.7660765647888184, "DECLINED COMMENT_f1": 0.4068688118811881, "DECLINED COMMENT_e": 0.5625, "PROPOSAL/ORDER/LAW_loss": 4.049454689025879, "PROPOSAL/ORDER/LAW_f1": 0.2889760348583878, "PROPOSAL/ORDER/LAW_e": 0.4266666666666667, "PRICE SIGNAL_loss": 3.616384744644165, "PRICE SIGNAL_f1": 0.4947368421052632, "PRICE SIGNAL_e": 0.5263157894736842, "NARRATIVE_loss": 1.9826314449310303, "NARRATIVE_f1": 0.6029803893362629, "NARRATIVE_e": 0.6985294117647058, "DIRECT OBSERVATION_loss": 5.71368932723999, "DIRECT OBSERVATION_f1": 0.01694915254237288, "DIRECT OBSERVATION_e": 0.2033898305084746, "COMMUNICATION_loss": 3.4060516357421875, "COMMUNICATION_f1": 0.5073826058201059, "COMMUNICATION_e": 0.5, "PUBLIC SPEECH_loss": 3.7426810264587402, "PUBLIC SPEECH_f1": 0.25891142160512753, "PUBLIC SPEECH_e": 0.4666666666666667, "VOTE/POLL_loss": 4.689935207366943, "VOTE/POLL_f1": 0.3312252964426877, "VOTE/POLL_e": 0.34782608695652173, "COURT PROCEEDING_loss": 5.710793972015381, "COURT PROCEEDING_f1": 0.2707763347763348, "COURT PROCEEDING_e": 0.44, "full_loss": 3.077216148376465, "full_f1": 0.5272328806773431, "full_e": 0.5724272930648769}

callback-metrics-state-20000.json

@@ -0,0 +1 @@
+{"INDIRECT QUOTE_loss": 2.1148886680603027, "INDIRECT QUOTE_f1": 0.6061409545975309, "INDIRECT QUOTE_e": 0.6154545454545455, "DIRECT QUOTE_loss": 2.783146381378174, "DIRECT QUOTE_f1": 0.5601386615724058, "DIRECT QUOTE_e": 0.6043225270157938, "BACKGROUND_loss": 3.3464651107788086, "BACKGROUND_f1": 0.6221739141179379, "BACKGROUND_e": 0.6682464454976303, "PUBLISHED WORK_loss": 3.9592363834381104, "PUBLISHED WORK_f1": 0.43717210437759113, "PUBLISHED WORK_e": 0.5038167938931297, "STATEMENT_loss": 2.1502439975738525, "STATEMENT_f1": 0.5222579773709637, "STATEMENT_e": 0.6381578947368421, "SOCIAL MEDIA POST_loss": 4.850955963134766, "SOCIAL MEDIA POST_f1": 0.15151515151515152, "SOCIAL MEDIA POST_e": 0.2727272727272727, "PRESS REPORT_loss": 4.060721397399902, "PRESS REPORT_f1": 0.45439298174801934, "PRESS REPORT_e": 0.5, "DECLINED COMMENT_loss": 3.9542932510375977, "DECLINED COMMENT_f1": 0.375, "DECLINED COMMENT_e": 0.5, "PROPOSAL/ORDER/LAW_loss": 3.2367727756500244, "PROPOSAL/ORDER/LAW_f1": 0.32273568107458245, "PROPOSAL/ORDER/LAW_e": 0.44, "PRICE SIGNAL_loss": 3.1010141372680664, "PRICE SIGNAL_f1": 0.45087719298245615, "PRICE SIGNAL_e": 0.47368421052631576, "NARRATIVE_loss": 1.8089553117752075, "NARRATIVE_f1": 0.6354846971078675, "NARRATIVE_e": 0.7352941176470589, "DIRECT OBSERVATION_loss": 3.868048906326294, "DIRECT OBSERVATION_f1": 0.01694915254237288, "DIRECT OBSERVATION_e": 0.4067796610169492, "COMMUNICATION_loss": 3.3353352546691895, "COMMUNICATION_f1": 0.570775462962963, "COMMUNICATION_e": 0.5625, "PUBLIC SPEECH_loss": 3.9616973400115967, "PUBLIC SPEECH_f1": 0.3, "PUBLIC SPEECH_e": 0.5666666666666667, "VOTE/POLL_loss": 4.4696855545043945, "VOTE/POLL_f1": 0.2924901185770751, "VOTE/POLL_e": 0.30434782608695654, "COURT PROCEEDING_loss": 5.168118953704834, "COURT PROCEEDING_f1": 0.34277633477633473, "COURT PROCEEDING_e": 0.44, "full_loss": 2.754518747329712, "full_f1": 0.5501093051292324, "full_e": 0.6009507829977628}

callback-metrics-state-5000.json

@@ -0,0 +1 @@
+{"INDIRECT QUOTE_loss": 2.8530545234680176, "INDIRECT QUOTE_f1": 0.4941323500870946, "INDIRECT QUOTE_e": 0.5, "DIRECT QUOTE_loss": 3.684295892715454, "DIRECT QUOTE_f1": 0.42220711015100093, "DIRECT QUOTE_e": 0.4513715710723192, "BACKGROUND_loss": 3.72399640083313, "BACKGROUND_f1": 0.5459073471062601, "BACKGROUND_e": 0.5829383886255924, "PUBLISHED WORK_loss": 5.283926010131836, "PUBLISHED WORK_f1": 0.33880863285241525, "PUBLISHED WORK_e": 0.37404580152671757, "STATEMENT_loss": 3.947087287902832, "STATEMENT_f1": 0.40469407680490305, "STATEMENT_e": 0.4934210526315789, "SOCIAL MEDIA POST_loss": 6.421342372894287, "SOCIAL MEDIA POST_f1": 0.09113693324219639, "SOCIAL MEDIA POST_e": 0.18181818181818182, "PRESS REPORT_loss": 7.026741981506348, "PRESS REPORT_f1": 0.2708760847470401, "PRESS REPORT_e": 0.27586206896551724, "DECLINED COMMENT_loss": 6.057079315185547, "DECLINED COMMENT_f1": 0.34375, "DECLINED COMMENT_e": 0.4375, "PROPOSAL/ORDER/LAW_loss": 6.083311557769775, "PROPOSAL/ORDER/LAW_f1": 0.17483065083065083, "PROPOSAL/ORDER/LAW_e": 0.24, "PRICE SIGNAL_loss": 5.802990913391113, "PRICE SIGNAL_f1": 0.3157894736842105, "PRICE SIGNAL_e": 0.3684210526315789, "NARRATIVE_loss": 2.563370704650879, "NARRATIVE_f1": 0.560518411621996, "NARRATIVE_e": 0.6397058823529411, "DIRECT OBSERVATION_loss": 7.233707904815674, "DIRECT OBSERVATION_f1": 0.01694915254237288, "DIRECT OBSERVATION_e": 0.01694915254237288, "COMMUNICATION_loss": 5.04656982421875, "COMMUNICATION_f1": 0.4015110848372835, "COMMUNICATION_e": 0.40625, "PUBLIC SPEECH_loss": 6.0680975914001465, "PUBLIC SPEECH_f1": 0.255, "PUBLIC SPEECH_e": 0.4666666666666667, "VOTE/POLL_loss": 7.1436967849731445, "VOTE/POLL_f1": 0.26600790513833994, "VOTE/POLL_e": 0.2608695652173913, "COURT PROCEEDING_loss": 6.0425310134887695, "COURT PROCEEDING_f1": 0.21535511250254696, "COURT PROCEEDING_e": 0.4, "full_loss": 3.77375864982605, "full_f1": 0.4364628001202054, "full_e": 0.4684004474272931}

config.json

@@ -0,0 +1,38 @@
+{
+  "_name_or_path": "google/bigbird-roberta-base",
+  "architectures": [
+    "QAModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "attention_type": "original_full",
+  "block_size": 64,
+  "bos_token_id": 1,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "freeze_layers": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu_new",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "include_nones_as_positives": false,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "loss_window": null,
+  "max_position_embeddings": 4096,
+  "model_type": "big_bird",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_random_blocks": 3,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "qa_head": {},
+  "rescale_embeddings": false,
+  "sep_token_id": 66,
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.2",
+  "type_vocab_size": 2,
+  "use_bias": true,
+  "use_cache": true,
+  "vocab_size": 50358
+}

post-training eval_results.json

@@ -0,0 +1,9 @@
+{
+    "eval_samples": 3576,
+    "test_e": 0.6023489932885906,
+    "test_f1": 0.5506494789202888,
+    "test_loss": 2.7486965656280518,
+    "test_runtime": 80.778,
+    "test_samples_per_second": 44.269,
+    "test_steps_per_second": 44.269
+}

pre-training eval_results.json

@@ -0,0 +1,8 @@
+{
+    "eval_e": 0.0,
+    "eval_f1": 0.003218218203323381,
+    "eval_loss": 7.38798189163208,
+    "eval_runtime": 81.0639,
+    "eval_samples_per_second": 44.113,
+    "eval_steps_per_second": 44.113
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fadb5be0bb0e164763126f1767ffd20cdbb87d2bf3b4104c66e5ed5af18d2d43
+size 509994093

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.0,
+    "train_loss": 1.5977188166040575,
+    "train_runtime": 2455.1123,
+    "train_samples": 10138,
+    "train_samples_per_second": 8.259,
+    "train_steps_per_second": 8.259
+}

trainer_state.json

@@ -0,0 +1,305 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 20276,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8767015190372854e-05,
+      "loss": 3.6543,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7534030380745706e-05,
+      "loss": 2.9884,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.630104557111857e-05,
+      "loss": 2.6586,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.506806076149142e-05,
+      "loss": 2.6187,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3835075951864274e-05,
+      "loss": 2.5105,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2602091142237125e-05,
+      "loss": 2.235,
+      "step": 3000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1369106332609984e-05,
+      "loss": 2.2836,
+      "step": 3500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.013612152298284e-05,
+      "loss": 2.1538,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.8903136713355694e-05,
+      "loss": 2.2132,
+      "step": 4500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7670151903728545e-05,
+      "loss": 2.1561,
+      "step": 5000
+    },
+    {
+      "epoch": 0.49,
+      "eval_e": 0.4684004474272931,
+      "eval_f1": 0.4364628001202054,
+      "eval_loss": 3.77375864982605,
+      "eval_runtime": 80.8413,
+      "eval_samples_per_second": 44.235,
+      "eval_steps_per_second": 44.235,
+      "step": 5000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6437167094101404e-05,
+      "loss": 2.0688,
+      "step": 5500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5204182284474255e-05,
+      "loss": 2.0712,
+      "step": 6000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3971197474847113e-05,
+      "loss": 1.7271,
+      "step": 6500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2738212665219965e-05,
+      "loss": 1.7146,
+      "step": 7000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.150522785559282e-05,
+      "loss": 1.8106,
+      "step": 7500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0272243045965675e-05,
+      "loss": 1.7478,
+      "step": 8000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.903925823633853e-05,
+      "loss": 1.8638,
+      "step": 8500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.780627342671138e-05,
+      "loss": 1.41,
+      "step": 9000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.657328861708424e-05,
+      "loss": 1.7941,
+      "step": 9500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5340303807457095e-05,
+      "loss": 1.5219,
+      "step": 10000
+    },
+    {
+      "epoch": 0.99,
+      "eval_e": 0.5542505592841164,
+      "eval_f1": 0.5065431932934646,
+      "eval_loss": 2.430802822113037,
+      "eval_runtime": 80.8808,
+      "eval_samples_per_second": 44.213,
+      "eval_steps_per_second": 44.213,
+      "step": 10000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4107318997829946e-05,
+      "loss": 1.2346,
+      "step": 10500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.2874334188202805e-05,
+      "loss": 1.2315,
+      "step": 11000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.1641349378575656e-05,
+      "loss": 1.2655,
+      "step": 11500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.040836456894851e-05,
+      "loss": 1.2839,
+      "step": 12000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9175379759321366e-05,
+      "loss": 1.0951,
+      "step": 12500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.794239494969422e-05,
+      "loss": 1.2722,
+      "step": 13000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.6709410140067076e-05,
+      "loss": 1.0126,
+      "step": 13500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.547642533043993e-05,
+      "loss": 1.1157,
+      "step": 14000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4243440520812784e-05,
+      "loss": 1.1724,
+      "step": 14500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.301045571118564e-05,
+      "loss": 0.9838,
+      "step": 15000
+    },
+    {
+      "epoch": 1.48,
+      "eval_e": 0.5724272930648769,
+      "eval_f1": 0.5272328806773431,
+      "eval_loss": 3.077216148376465,
+      "eval_runtime": 81.057,
+      "eval_samples_per_second": 44.117,
+      "eval_steps_per_second": 44.117,
+      "step": 15000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1777470901558493e-05,
+      "loss": 0.9351,
+      "step": 15500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0544486091931348e-05,
+      "loss": 0.997,
+      "step": 16000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.311501282304202e-06,
+      "loss": 0.9782,
+      "step": 16500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.078516472677057e-06,
+      "loss": 0.9934,
+      "step": 17000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.8455316630499115e-06,
+      "loss": 0.9395,
+      "step": 17500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.6125468534227665e-06,
+      "loss": 0.8001,
+      "step": 18000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.379562043795621e-06,
+      "loss": 0.9129,
+      "step": 18500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1465772341684756e-06,
+      "loss": 1.0251,
+      "step": 19000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9135924245413297e-06,
+      "loss": 1.0114,
+      "step": 19500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.806076149141843e-07,
+      "loss": 0.8334,
+      "step": 20000
+    },
+    {
+      "epoch": 1.97,
+      "eval_e": 0.6009507829977628,
+      "eval_f1": 0.5501093051292324,
+      "eval_loss": 2.754518747329712,
+      "eval_runtime": 81.0315,
+      "eval_samples_per_second": 44.131,
+      "eval_steps_per_second": 44.131,
+      "step": 20000
+    },
+    {
+      "epoch": 2.0,
+      "step": 20276,
+      "total_flos": 1.023741888632904e+16,
+      "train_loss": 1.5977188166040575,
+      "train_runtime": 2455.1123,
+      "train_samples_per_second": 8.259,
+      "train_steps_per_second": 8.259
+    }
+  ],
+  "max_steps": 20276,
+  "num_train_epochs": 2,
+  "total_flos": 1.023741888632904e+16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:188c916efd0aab3620a4aedbf5879574a997104213f4553d6b96553c56e32c9c
+size 2927