Upload folder using huggingface_hub

README.md

@@ -0,0 +1,25 @@
+---
+license: apache-2.0
+language: en
+tags:
+- token-classification
+- named-entity-recognition
+- roberta
+- crf
+- news
+---
+
+# roberta-crf-boundary-news-ner
+
+RoBERTa + CRF news NER model, extended from [`Souvikbasur/roberta-crf-news-ner`]({"https://huggingface.co/" + CHECKPOINT_REPO}) with:
+
+- A boundary-guided interaction module (auxiliary start/end token classifiers gating the shared representation before the CRF classifier), inspired by Zheng et al. (2019, EMNLP-IJCNLP) and Gao et al. (2022).
+- Boundary-smoothed multi-task loss (Zhu & Li, 2022, ACL) combining the CRF negative log-likelihood with the boundary auxiliary loss.
+- FGM adversarial training on the token embeddings (Miyato et al., 2017).
+- An optional confidence-based boundary-correction post-processing step at inference.
+
+**This is a fine-tuning extension of an existing checkpoint, evaluated with an ablation against the base CRF model — see the linked paper/repo for the actual measured numbers.** Do not assume improvement without checking the evaluation cells in the training notebook.
+
+Trained on the same 17-category custom news NER dataset and train/val/test split as `Souvikbasur/roberta-crf-news-ner`.
+
+Requires the custom `RobertaCRFBoundaryForNER` class (see the training notebook) to load — a bare `state_dict` isn't loadable via `AutoModel.from_pretrained` alone.

checkpoint-1218/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71ddfbd947fd40df5bd83d2e498342a7472c87570aba77012ea062fd73ea35dc
+size 498726232

checkpoint-1218/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f6ef20c4c94da24dd28b6f8ee6863a03d980527568887ff647c477d6978452c
+size 992853771

checkpoint-1218/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3333aa4f4aa9deaf4dff794bba6ebdd4ea7059d52bfe4ee04d4fb75985605907
+size 14645

checkpoint-1218/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bd4703dfcdfd846963d8e4c44639fa2977608e55b62072748ed7f967af92e45
+size 1383

checkpoint-1218/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e72d55934b5e6c9f237d14aae791f6881a825893c3941c3bb2f50f5cf78c044d
+size 1465

checkpoint-1218/trainer_state.json

@@ -0,0 +1,233 @@
+{
+  "best_global_step": 1218,
+  "best_metric": 0.8758669056339127,
+  "best_model_checkpoint": "./ner_crf_boundary_output/checkpoint-1218",
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1218,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08213552361396304,
+      "grad_norm": 289.9013977050781,
+      "learning_rate": 2.677595628415301e-06,
+      "loss": 7.03633056640625,
+      "step": 50
+    },
+    {
+      "epoch": 0.16427104722792607,
+      "grad_norm": 397.5795593261719,
+      "learning_rate": 5.409836065573772e-06,
+      "loss": 5.944567260742187,
+      "step": 100
+    },
+    {
+      "epoch": 0.2464065708418891,
+      "grad_norm": 1030.340576171875,
+      "learning_rate": 8.14207650273224e-06,
+      "loss": 6.386118774414062,
+      "step": 150
+    },
+    {
+      "epoch": 0.32854209445585214,
+      "grad_norm": 416.71173095703125,
+      "learning_rate": 9.944095038434662e-06,
+      "loss": 6.638558349609375,
+      "step": 200
+    },
+    {
+      "epoch": 0.4106776180698152,
+      "grad_norm": 379.8328857421875,
+      "learning_rate": 9.769392033542979e-06,
+      "loss": 5.887216186523437,
+      "step": 250
+    },
+    {
+      "epoch": 0.4928131416837782,
+      "grad_norm": 328.294921875,
+      "learning_rate": 9.594689028651294e-06,
+      "loss": 6.057736206054687,
+      "step": 300
+    },
+    {
+      "epoch": 0.5749486652977412,
+      "grad_norm": 306.49560546875,
+      "learning_rate": 9.41998602375961e-06,
+      "loss": 6.300731201171875,
+      "step": 350
+    },
+    {
+      "epoch": 0.6570841889117043,
+      "grad_norm": 406.1339111328125,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 5.523515625,
+      "step": 400
+    },
+    {
+      "epoch": 0.7392197125256673,
+      "grad_norm": 405.34136962890625,
+      "learning_rate": 9.070580013976241e-06,
+      "loss": 6.377322998046875,
+      "step": 450
+    },
+    {
+      "epoch": 0.8213552361396304,
+      "grad_norm": 219.61463928222656,
+      "learning_rate": 8.895877009084557e-06,
+      "loss": 5.774469604492188,
+      "step": 500
+    },
+    {
+      "epoch": 0.9034907597535934,
+      "grad_norm": 294.69683837890625,
+      "learning_rate": 8.721174004192873e-06,
+      "loss": 6.709805297851562,
+      "step": 550
+    },
+    {
+      "epoch": 0.9856262833675564,
+      "grad_norm": 397.64190673828125,
+      "learning_rate": 8.546470999301189e-06,
+      "loss": 5.980140991210938,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.8755947274003588,
+      "eval_loss": 4.1545538902282715,
+      "eval_precision": 0.8619471744471745,
+      "eval_recall": 0.8896814075130766,
+      "eval_runtime": 44.15,
+      "eval_samples_per_second": 49.015,
+      "eval_steps_per_second": 6.138,
+      "step": 609
+    },
+    {
+      "epoch": 1.0673511293634497,
+      "grad_norm": 406.8649597167969,
+      "learning_rate": 8.371767994409504e-06,
+      "loss": 5.502258911132812,
+      "step": 650
+    },
+    {
+      "epoch": 1.1494866529774128,
+      "grad_norm": 353.2889709472656,
+      "learning_rate": 8.19706498951782e-06,
+      "loss": 5.161415405273438,
+      "step": 700
+    },
+    {
+      "epoch": 1.2316221765913757,
+      "grad_norm": 524.7266845703125,
+      "learning_rate": 8.022361984626136e-06,
+      "loss": 6.095115966796875,
+      "step": 750
+    },
+    {
+      "epoch": 1.3137577002053389,
+      "grad_norm": 532.918212890625,
+      "learning_rate": 7.847658979734451e-06,
+      "loss": 5.310472412109375,
+      "step": 800
+    },
+    {
+      "epoch": 1.3958932238193018,
+      "grad_norm": 481.7142639160156,
+      "learning_rate": 7.672955974842768e-06,
+      "loss": 5.16358154296875,
+      "step": 850
+    },
+    {
+      "epoch": 1.478028747433265,
+      "grad_norm": 427.3961181640625,
+      "learning_rate": 7.498252969951084e-06,
+      "loss": 5.792083740234375,
+      "step": 900
+    },
+    {
+      "epoch": 1.5601642710472279,
+      "grad_norm": 571.2908935546875,
+      "learning_rate": 7.3235499650594e-06,
+      "loss": 4.886295166015625,
+      "step": 950
+    },
+    {
+      "epoch": 1.642299794661191,
+      "grad_norm": 310.74542236328125,
+      "learning_rate": 7.148846960167716e-06,
+      "loss": 5.547396240234375,
+      "step": 1000
+    },
+    {
+      "epoch": 1.724435318275154,
+      "grad_norm": 210.18771362304688,
+      "learning_rate": 6.974143955276031e-06,
+      "loss": 5.537097778320312,
+      "step": 1050
+    },
+    {
+      "epoch": 1.806570841889117,
+      "grad_norm": 323.7012023925781,
+      "learning_rate": 6.799440950384347e-06,
+      "loss": 4.974610290527344,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8887063655030802,
+      "grad_norm": 300.61309814453125,
+      "learning_rate": 6.624737945492663e-06,
+      "loss": 5.379744262695312,
+      "step": 1150
+    },
+    {
+      "epoch": 1.9708418891170432,
+      "grad_norm": 420.67388916015625,
+      "learning_rate": 6.4500349406009785e-06,
+      "loss": 5.358375854492188,
+      "step": 1200
+    },
+    {
+      "epoch": 2.0,
+      "eval_f1": 0.8758669056339127,
+      "eval_loss": 4.113088607788086,
+      "eval_precision": 0.8614347026364194,
+      "eval_recall": 0.8907909335869393,
+      "eval_runtime": 44.1674,
+      "eval_samples_per_second": 48.995,
+      "eval_steps_per_second": 6.136,
+      "step": 1218
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 3045,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 2,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "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": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1218/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97858c4638f99430a1879ac2e22898be13fdaf4eb03fabedc400f472205c63d0
+size 5201

checkpoint-1827/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a17b1906bf894de25967d6a2d5e15df812498b969a62a11fbfb153ea94ee26d
+size 498726232

checkpoint-1827/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e504117b09fd3181875cc445dc2457360427d46341da471cbafc51ae8d7d295
+size 992853771

checkpoint-1827/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e44b998b6b0197c7da62669120aade6212bfda5418f8287238653bed219218f
+size 14645

checkpoint-1827/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08068d73678cb958af4419d7cbdc9fa2c0f3499a4a0709124ee2427e5f4fb22f
+size 1383

checkpoint-1827/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:383c9d1981e6e21e2d82c6d27133f03456caa6d4ed7309a2e244cf88d5406c2b
+size 1465

checkpoint-1827/trainer_state.json

@@ -0,0 +1,328 @@
+{
+  "best_global_step": 1827,
+  "best_metric": 0.88046875,
+  "best_model_checkpoint": "./ner_crf_boundary_output/checkpoint-1827",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1827,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08213552361396304,
+      "grad_norm": 289.9013977050781,
+      "learning_rate": 2.677595628415301e-06,
+      "loss": 7.03633056640625,
+      "step": 50
+    },
+    {
+      "epoch": 0.16427104722792607,
+      "grad_norm": 397.5795593261719,
+      "learning_rate": 5.409836065573772e-06,
+      "loss": 5.944567260742187,
+      "step": 100
+    },
+    {
+      "epoch": 0.2464065708418891,
+      "grad_norm": 1030.340576171875,
+      "learning_rate": 8.14207650273224e-06,
+      "loss": 6.386118774414062,
+      "step": 150
+    },
+    {
+      "epoch": 0.32854209445585214,
+      "grad_norm": 416.71173095703125,
+      "learning_rate": 9.944095038434662e-06,
+      "loss": 6.638558349609375,
+      "step": 200
+    },
+    {
+      "epoch": 0.4106776180698152,
+      "grad_norm": 379.8328857421875,
+      "learning_rate": 9.769392033542979e-06,
+      "loss": 5.887216186523437,
+      "step": 250
+    },
+    {
+      "epoch": 0.4928131416837782,
+      "grad_norm": 328.294921875,
+      "learning_rate": 9.594689028651294e-06,
+      "loss": 6.057736206054687,
+      "step": 300
+    },
+    {
+      "epoch": 0.5749486652977412,
+      "grad_norm": 306.49560546875,
+      "learning_rate": 9.41998602375961e-06,
+      "loss": 6.300731201171875,
+      "step": 350
+    },
+    {
+      "epoch": 0.6570841889117043,
+      "grad_norm": 406.1339111328125,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 5.523515625,
+      "step": 400
+    },
+    {
+      "epoch": 0.7392197125256673,
+      "grad_norm": 405.34136962890625,
+      "learning_rate": 9.070580013976241e-06,
+      "loss": 6.377322998046875,
+      "step": 450
+    },
+    {
+      "epoch": 0.8213552361396304,
+      "grad_norm": 219.61463928222656,
+      "learning_rate": 8.895877009084557e-06,
+      "loss": 5.774469604492188,
+      "step": 500
+    },
+    {
+      "epoch": 0.9034907597535934,
+      "grad_norm": 294.69683837890625,
+      "learning_rate": 8.721174004192873e-06,
+      "loss": 6.709805297851562,
+      "step": 550
+    },
+    {
+      "epoch": 0.9856262833675564,
+      "grad_norm": 397.64190673828125,
+      "learning_rate": 8.546470999301189e-06,
+      "loss": 5.980140991210938,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.8755947274003588,
+      "eval_loss": 4.1545538902282715,
+      "eval_precision": 0.8619471744471745,
+      "eval_recall": 0.8896814075130766,
+      "eval_runtime": 44.15,
+      "eval_samples_per_second": 49.015,
+      "eval_steps_per_second": 6.138,
+      "step": 609
+    },
+    {
+      "epoch": 1.0673511293634497,
+      "grad_norm": 406.8649597167969,
+      "learning_rate": 8.371767994409504e-06,
+      "loss": 5.502258911132812,
+      "step": 650
+    },
+    {
+      "epoch": 1.1494866529774128,
+      "grad_norm": 353.2889709472656,
+      "learning_rate": 8.19706498951782e-06,
+      "loss": 5.161415405273438,
+      "step": 700
+    },
+    {
+      "epoch": 1.2316221765913757,
+      "grad_norm": 524.7266845703125,
+      "learning_rate": 8.022361984626136e-06,
+      "loss": 6.095115966796875,
+      "step": 750
+    },
+    {
+      "epoch": 1.3137577002053389,
+      "grad_norm": 532.918212890625,
+      "learning_rate": 7.847658979734451e-06,
+      "loss": 5.310472412109375,
+      "step": 800
+    },
+    {
+      "epoch": 1.3958932238193018,
+      "grad_norm": 481.7142639160156,
+      "learning_rate": 7.672955974842768e-06,
+      "loss": 5.16358154296875,
+      "step": 850
+    },
+    {
+      "epoch": 1.478028747433265,
+      "grad_norm": 427.3961181640625,
+      "learning_rate": 7.498252969951084e-06,
+      "loss": 5.792083740234375,
+      "step": 900
+    },
+    {
+      "epoch": 1.5601642710472279,
+      "grad_norm": 571.2908935546875,
+      "learning_rate": 7.3235499650594e-06,
+      "loss": 4.886295166015625,
+      "step": 950
+    },
+    {
+      "epoch": 1.642299794661191,
+      "grad_norm": 310.74542236328125,
+      "learning_rate": 7.148846960167716e-06,
+      "loss": 5.547396240234375,
+      "step": 1000
+    },
+    {
+      "epoch": 1.724435318275154,
+      "grad_norm": 210.18771362304688,
+      "learning_rate": 6.974143955276031e-06,
+      "loss": 5.537097778320312,
+      "step": 1050
+    },
+    {
+      "epoch": 1.806570841889117,
+      "grad_norm": 323.7012023925781,
+      "learning_rate": 6.799440950384347e-06,
+      "loss": 4.974610290527344,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8887063655030802,
+      "grad_norm": 300.61309814453125,
+      "learning_rate": 6.624737945492663e-06,
+      "loss": 5.379744262695312,
+      "step": 1150
+    },
+    {
+      "epoch": 1.9708418891170432,
+      "grad_norm": 420.67388916015625,
+      "learning_rate": 6.4500349406009785e-06,
+      "loss": 5.358375854492188,
+      "step": 1200
+    },
+    {
+      "epoch": 2.0,
+      "eval_f1": 0.8758669056339127,
+      "eval_loss": 4.113088607788086,
+      "eval_precision": 0.8614347026364194,
+      "eval_recall": 0.8907909335869393,
+      "eval_runtime": 44.1674,
+      "eval_samples_per_second": 48.995,
+      "eval_steps_per_second": 6.136,
+      "step": 1218
+    },
+    {
+      "epoch": 2.0525667351129364,
+      "grad_norm": 716.19775390625,
+      "learning_rate": 6.275331935709295e-06,
+      "loss": 5.002279052734375,
+      "step": 1250
+    },
+    {
+      "epoch": 2.1347022587268993,
+      "grad_norm": 269.1075439453125,
+      "learning_rate": 6.100628930817611e-06,
+      "loss": 4.724185180664063,
+      "step": 1300
+    },
+    {
+      "epoch": 2.2168377823408623,
+      "grad_norm": 386.7925720214844,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 4.566873779296875,
+      "step": 1350
+    },
+    {
+      "epoch": 2.2989733059548256,
+      "grad_norm": 348.7635803222656,
+      "learning_rate": 5.751222921034242e-06,
+      "loss": 4.5791323852539065,
+      "step": 1400
+    },
+    {
+      "epoch": 2.3811088295687886,
+      "grad_norm": 226.85601806640625,
+      "learning_rate": 5.576519916142558e-06,
+      "loss": 4.002293090820313,
+      "step": 1450
+    },
+    {
+      "epoch": 2.4632443531827515,
+      "grad_norm": 324.4167785644531,
+      "learning_rate": 5.401816911250875e-06,
+      "loss": 4.811328125,
+      "step": 1500
+    },
+    {
+      "epoch": 2.545379876796715,
+      "grad_norm": 265.19696044921875,
+      "learning_rate": 5.22711390635919e-06,
+      "loss": 4.3527978515625,
+      "step": 1550
+    },
+    {
+      "epoch": 2.6275154004106778,
+      "grad_norm": 255.17453002929688,
+      "learning_rate": 5.052410901467506e-06,
+      "loss": 4.515827026367187,
+      "step": 1600
+    },
+    {
+      "epoch": 2.7096509240246407,
+      "grad_norm": 343.71630859375,
+      "learning_rate": 4.877707896575822e-06,
+      "loss": 4.646475830078125,
+      "step": 1650
+    },
+    {
+      "epoch": 2.7917864476386036,
+      "grad_norm": 309.7668151855469,
+      "learning_rate": 4.703004891684137e-06,
+      "loss": 4.717439880371094,
+      "step": 1700
+    },
+    {
+      "epoch": 2.8739219712525665,
+      "grad_norm": 211.75811767578125,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 4.192425231933594,
+      "step": 1750
+    },
+    {
+      "epoch": 2.95605749486653,
+      "grad_norm": 471.5155334472656,
+      "learning_rate": 4.353598881900769e-06,
+      "loss": 5.051603088378906,
+      "step": 1800
+    },
+    {
+      "epoch": 3.0,
+      "eval_f1": 0.88046875,
+      "eval_loss": 4.196066379547119,
+      "eval_precision": 0.8681250962871668,
+      "eval_recall": 0.8931684894595023,
+      "eval_runtime": 43.3428,
+      "eval_samples_per_second": 49.928,
+      "eval_steps_per_second": 6.252,
+      "step": 1827
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 3045,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 2,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "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": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1827/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97858c4638f99430a1879ac2e22898be13fdaf4eb03fabedc400f472205c63d0
+size 5201

checkpoint-2436/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50811bde28c596c4fbe5db19dcacf283afd6b2eae1a1fea487b75884b1fbb639
+size 498726232

checkpoint-2436/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84844e11d8b03e1078e0ffdd0774f72a32a9162c636730a1bbeebd544863b07f
+size 992853771

checkpoint-2436/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:569718648fb60fab03c21d94977589bf9498a468484b287e3b0512152efabade
+size 14645

checkpoint-2436/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17c665e498639ee7cc715d7270d31dae11cbf99ee006620ed8bb359912f16839
+size 1383

checkpoint-2436/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9773c117cd3c0c029bd979b5ab4e8cca7d2b48abdd33e7db2fda1da3ad39da3
+size 1465

checkpoint-2436/trainer_state.json

@@ -0,0 +1,423 @@
+{
+  "best_global_step": 1827,
+  "best_metric": 0.88046875,
+  "best_model_checkpoint": "./ner_crf_boundary_output/checkpoint-1827",
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 2436,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08213552361396304,
+      "grad_norm": 289.9013977050781,
+      "learning_rate": 2.677595628415301e-06,
+      "loss": 7.03633056640625,
+      "step": 50
+    },
+    {
+      "epoch": 0.16427104722792607,
+      "grad_norm": 397.5795593261719,
+      "learning_rate": 5.409836065573772e-06,
+      "loss": 5.944567260742187,
+      "step": 100
+    },
+    {
+      "epoch": 0.2464065708418891,
+      "grad_norm": 1030.340576171875,
+      "learning_rate": 8.14207650273224e-06,
+      "loss": 6.386118774414062,
+      "step": 150
+    },
+    {
+      "epoch": 0.32854209445585214,
+      "grad_norm": 416.71173095703125,
+      "learning_rate": 9.944095038434662e-06,
+      "loss": 6.638558349609375,
+      "step": 200
+    },
+    {
+      "epoch": 0.4106776180698152,
+      "grad_norm": 379.8328857421875,
+      "learning_rate": 9.769392033542979e-06,
+      "loss": 5.887216186523437,
+      "step": 250
+    },
+    {
+      "epoch": 0.4928131416837782,
+      "grad_norm": 328.294921875,
+      "learning_rate": 9.594689028651294e-06,
+      "loss": 6.057736206054687,
+      "step": 300
+    },
+    {
+      "epoch": 0.5749486652977412,
+      "grad_norm": 306.49560546875,
+      "learning_rate": 9.41998602375961e-06,
+      "loss": 6.300731201171875,
+      "step": 350
+    },
+    {
+      "epoch": 0.6570841889117043,
+      "grad_norm": 406.1339111328125,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 5.523515625,
+      "step": 400
+    },
+    {
+      "epoch": 0.7392197125256673,
+      "grad_norm": 405.34136962890625,
+      "learning_rate": 9.070580013976241e-06,
+      "loss": 6.377322998046875,
+      "step": 450
+    },
+    {
+      "epoch": 0.8213552361396304,
+      "grad_norm": 219.61463928222656,
+      "learning_rate": 8.895877009084557e-06,
+      "loss": 5.774469604492188,
+      "step": 500
+    },
+    {
+      "epoch": 0.9034907597535934,
+      "grad_norm": 294.69683837890625,
+      "learning_rate": 8.721174004192873e-06,
+      "loss": 6.709805297851562,
+      "step": 550
+    },
+    {
+      "epoch": 0.9856262833675564,
+      "grad_norm": 397.64190673828125,
+      "learning_rate": 8.546470999301189e-06,
+      "loss": 5.980140991210938,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.8755947274003588,
+      "eval_loss": 4.1545538902282715,
+      "eval_precision": 0.8619471744471745,
+      "eval_recall": 0.8896814075130766,
+      "eval_runtime": 44.15,
+      "eval_samples_per_second": 49.015,
+      "eval_steps_per_second": 6.138,
+      "step": 609
+    },
+    {
+      "epoch": 1.0673511293634497,
+      "grad_norm": 406.8649597167969,
+      "learning_rate": 8.371767994409504e-06,
+      "loss": 5.502258911132812,
+      "step": 650
+    },
+    {
+      "epoch": 1.1494866529774128,
+      "grad_norm": 353.2889709472656,
+      "learning_rate": 8.19706498951782e-06,
+      "loss": 5.161415405273438,
+      "step": 700
+    },
+    {
+      "epoch": 1.2316221765913757,
+      "grad_norm": 524.7266845703125,
+      "learning_rate": 8.022361984626136e-06,
+      "loss": 6.095115966796875,
+      "step": 750
+    },
+    {
+      "epoch": 1.3137577002053389,
+      "grad_norm": 532.918212890625,
+      "learning_rate": 7.847658979734451e-06,
+      "loss": 5.310472412109375,
+      "step": 800
+    },
+    {
+      "epoch": 1.3958932238193018,
+      "grad_norm": 481.7142639160156,
+      "learning_rate": 7.672955974842768e-06,
+      "loss": 5.16358154296875,
+      "step": 850
+    },
+    {
+      "epoch": 1.478028747433265,
+      "grad_norm": 427.3961181640625,
+      "learning_rate": 7.498252969951084e-06,
+      "loss": 5.792083740234375,
+      "step": 900
+    },
+    {
+      "epoch": 1.5601642710472279,
+      "grad_norm": 571.2908935546875,
+      "learning_rate": 7.3235499650594e-06,
+      "loss": 4.886295166015625,
+      "step": 950
+    },
+    {
+      "epoch": 1.642299794661191,
+      "grad_norm": 310.74542236328125,
+      "learning_rate": 7.148846960167716e-06,
+      "loss": 5.547396240234375,
+      "step": 1000
+    },
+    {
+      "epoch": 1.724435318275154,
+      "grad_norm": 210.18771362304688,
+      "learning_rate": 6.974143955276031e-06,
+      "loss": 5.537097778320312,
+      "step": 1050
+    },
+    {
+      "epoch": 1.806570841889117,
+      "grad_norm": 323.7012023925781,
+      "learning_rate": 6.799440950384347e-06,
+      "loss": 4.974610290527344,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8887063655030802,
+      "grad_norm": 300.61309814453125,
+      "learning_rate": 6.624737945492663e-06,
+      "loss": 5.379744262695312,
+      "step": 1150
+    },
+    {
+      "epoch": 1.9708418891170432,
+      "grad_norm": 420.67388916015625,
+      "learning_rate": 6.4500349406009785e-06,
+      "loss": 5.358375854492188,
+      "step": 1200
+    },
+    {
+      "epoch": 2.0,
+      "eval_f1": 0.8758669056339127,
+      "eval_loss": 4.113088607788086,
+      "eval_precision": 0.8614347026364194,
+      "eval_recall": 0.8907909335869393,
+      "eval_runtime": 44.1674,
+      "eval_samples_per_second": 48.995,
+      "eval_steps_per_second": 6.136,
+      "step": 1218
+    },
+    {
+      "epoch": 2.0525667351129364,
+      "grad_norm": 716.19775390625,
+      "learning_rate": 6.275331935709295e-06,
+      "loss": 5.002279052734375,
+      "step": 1250
+    },
+    {
+      "epoch": 2.1347022587268993,
+      "grad_norm": 269.1075439453125,
+      "learning_rate": 6.100628930817611e-06,
+      "loss": 4.724185180664063,
+      "step": 1300
+    },
+    {
+      "epoch": 2.2168377823408623,
+      "grad_norm": 386.7925720214844,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 4.566873779296875,
+      "step": 1350
+    },
+    {
+      "epoch": 2.2989733059548256,
+      "grad_norm": 348.7635803222656,
+      "learning_rate": 5.751222921034242e-06,
+      "loss": 4.5791323852539065,
+      "step": 1400
+    },
+    {
+      "epoch": 2.3811088295687886,
+      "grad_norm": 226.85601806640625,
+      "learning_rate": 5.576519916142558e-06,
+      "loss": 4.002293090820313,
+      "step": 1450
+    },
+    {
+      "epoch": 2.4632443531827515,
+      "grad_norm": 324.4167785644531,
+      "learning_rate": 5.401816911250875e-06,
+      "loss": 4.811328125,
+      "step": 1500
+    },
+    {
+      "epoch": 2.545379876796715,
+      "grad_norm": 265.19696044921875,
+      "learning_rate": 5.22711390635919e-06,
+      "loss": 4.3527978515625,
+      "step": 1550
+    },
+    {
+      "epoch": 2.6275154004106778,
+      "grad_norm": 255.17453002929688,
+      "learning_rate": 5.052410901467506e-06,
+      "loss": 4.515827026367187,
+      "step": 1600
+    },
+    {
+      "epoch": 2.7096509240246407,
+      "grad_norm": 343.71630859375,
+      "learning_rate": 4.877707896575822e-06,
+      "loss": 4.646475830078125,
+      "step": 1650
+    },
+    {
+      "epoch": 2.7917864476386036,
+      "grad_norm": 309.7668151855469,
+      "learning_rate": 4.703004891684137e-06,
+      "loss": 4.717439880371094,
+      "step": 1700
+    },
+    {
+      "epoch": 2.8739219712525665,
+      "grad_norm": 211.75811767578125,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 4.192425231933594,
+      "step": 1750
+    },
+    {
+      "epoch": 2.95605749486653,
+      "grad_norm": 471.5155334472656,
+      "learning_rate": 4.353598881900769e-06,
+      "loss": 5.051603088378906,
+      "step": 1800
+    },
+    {
+      "epoch": 3.0,
+      "eval_f1": 0.88046875,
+      "eval_loss": 4.196066379547119,
+      "eval_precision": 0.8681250962871668,
+      "eval_recall": 0.8931684894595023,
+      "eval_runtime": 43.3428,
+      "eval_samples_per_second": 49.928,
+      "eval_steps_per_second": 6.252,
+      "step": 1827
+    },
+    {
+      "epoch": 3.037782340862423,
+      "grad_norm": 235.75936889648438,
+      "learning_rate": 4.1788958770090845e-06,
+      "loss": 4.453041381835938,
+      "step": 1850
+    },
+    {
+      "epoch": 3.119917864476386,
+      "grad_norm": 343.1545715332031,
+      "learning_rate": 4.0041928721174006e-06,
+      "loss": 3.923203430175781,
+      "step": 1900
+    },
+    {
+      "epoch": 3.202053388090349,
+      "grad_norm": 272.7549743652344,
+      "learning_rate": 3.829489867225717e-06,
+      "loss": 4.473578796386719,
+      "step": 1950
+    },
+    {
+      "epoch": 3.284188911704312,
+      "grad_norm": 205.64874267578125,
+      "learning_rate": 3.6547868623340327e-06,
+      "loss": 3.830061340332031,
+      "step": 2000
+    },
+    {
+      "epoch": 3.3663244353182753,
+      "grad_norm": 508.5237121582031,
+      "learning_rate": 3.4800838574423483e-06,
+      "loss": 3.810235595703125,
+      "step": 2050
+    },
+    {
+      "epoch": 3.4484599589322382,
+      "grad_norm": 292.8867492675781,
+      "learning_rate": 3.3053808525506643e-06,
+      "loss": 3.886056213378906,
+      "step": 2100
+    },
+    {
+      "epoch": 3.530595482546201,
+      "grad_norm": 348.3034362792969,
+      "learning_rate": 3.13067784765898e-06,
+      "loss": 3.2952536010742186,
+      "step": 2150
+    },
+    {
+      "epoch": 3.612731006160164,
+      "grad_norm": 287.3987121582031,
+      "learning_rate": 2.9559748427672955e-06,
+      "loss": 4.358919067382812,
+      "step": 2200
+    },
+    {
+      "epoch": 3.6948665297741274,
+      "grad_norm": 186.44305419921875,
+      "learning_rate": 2.7812718378756116e-06,
+      "loss": 4.031048583984375,
+      "step": 2250
+    },
+    {
+      "epoch": 3.7770020533880904,
+      "grad_norm": 588.447509765625,
+      "learning_rate": 2.606568832983927e-06,
+      "loss": 4.055077819824219,
+      "step": 2300
+    },
+    {
+      "epoch": 3.8591375770020533,
+      "grad_norm": 305.3722229003906,
+      "learning_rate": 2.4318658280922432e-06,
+      "loss": 3.518572998046875,
+      "step": 2350
+    },
+    {
+      "epoch": 3.9412731006160167,
+      "grad_norm": 403.2993469238281,
+      "learning_rate": 2.2571628232005593e-06,
+      "loss": 4.462532958984375,
+      "step": 2400
+    },
+    {
+      "epoch": 4.0,
+      "eval_f1": 0.878969957081545,
+      "eval_loss": 4.369068145751953,
+      "eval_precision": 0.8656624654165386,
+      "eval_recall": 0.8926929782849897,
+      "eval_runtime": 42.2843,
+      "eval_samples_per_second": 51.177,
+      "eval_steps_per_second": 6.409,
+      "step": 2436
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 3045,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 2,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 1
+      }
+    },
+    "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": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-2436/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97858c4638f99430a1879ac2e22898be13fdaf4eb03fabedc400f472205c63d0
+size 5201

checkpoint-3045/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9adae0837bbdafbbf19af00e2797a7142e65e7e3405956b9d1a4c5eea15579e
+size 498726232

checkpoint-3045/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2e590ba247502d19c91f5a1d25facdf5c69a5d743ba6003a01152a4c0b5548d
+size 992853771

checkpoint-3045/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d5f2eb9435ee99ea4fe9b02b3ebc3860ac131a813471c7bfaecfd0cc620d080
+size 14645

checkpoint-3045/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34d0b0f543f1bb97d8e894591db9419003b133572ecd656f2c438b70a2164949
+size 1383

checkpoint-3045/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32b230e828c05430114849b04858aefb10e669ac7dd2dbd6c23ffda929dc4a56
+size 1465

checkpoint-3045/trainer_state.json

@@ -0,0 +1,518 @@
+{
+  "best_global_step": 1827,
+  "best_metric": 0.88046875,
+  "best_model_checkpoint": "./ner_crf_boundary_output/checkpoint-1827",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 3045,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08213552361396304,
+      "grad_norm": 289.9013977050781,
+      "learning_rate": 2.677595628415301e-06,
+      "loss": 7.03633056640625,
+      "step": 50
+    },
+    {
+      "epoch": 0.16427104722792607,
+      "grad_norm": 397.5795593261719,
+      "learning_rate": 5.409836065573772e-06,
+      "loss": 5.944567260742187,
+      "step": 100
+    },
+    {
+      "epoch": 0.2464065708418891,
+      "grad_norm": 1030.340576171875,
+      "learning_rate": 8.14207650273224e-06,
+      "loss": 6.386118774414062,
+      "step": 150
+    },
+    {
+      "epoch": 0.32854209445585214,
+      "grad_norm": 416.71173095703125,
+      "learning_rate": 9.944095038434662e-06,
+      "loss": 6.638558349609375,
+      "step": 200
+    },
+    {
+      "epoch": 0.4106776180698152,
+      "grad_norm": 379.8328857421875,
+      "learning_rate": 9.769392033542979e-06,
+      "loss": 5.887216186523437,
+      "step": 250
+    },
+    {
+      "epoch": 0.4928131416837782,
+      "grad_norm": 328.294921875,
+      "learning_rate": 9.594689028651294e-06,
+      "loss": 6.057736206054687,
+      "step": 300
+    },
+    {
+      "epoch": 0.5749486652977412,
+      "grad_norm": 306.49560546875,
+      "learning_rate": 9.41998602375961e-06,
+      "loss": 6.300731201171875,
+      "step": 350
+    },
+    {
+      "epoch": 0.6570841889117043,
+      "grad_norm": 406.1339111328125,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 5.523515625,
+      "step": 400
+    },
+    {
+      "epoch": 0.7392197125256673,
+      "grad_norm": 405.34136962890625,
+      "learning_rate": 9.070580013976241e-06,
+      "loss": 6.377322998046875,
+      "step": 450
+    },
+    {
+      "epoch": 0.8213552361396304,
+      "grad_norm": 219.61463928222656,
+      "learning_rate": 8.895877009084557e-06,
+      "loss": 5.774469604492188,
+      "step": 500
+    },
+    {
+      "epoch": 0.9034907597535934,
+      "grad_norm": 294.69683837890625,
+      "learning_rate": 8.721174004192873e-06,
+      "loss": 6.709805297851562,
+      "step": 550
+    },
+    {
+      "epoch": 0.9856262833675564,
+      "grad_norm": 397.64190673828125,
+      "learning_rate": 8.546470999301189e-06,
+      "loss": 5.980140991210938,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.8755947274003588,
+      "eval_loss": 4.1545538902282715,
+      "eval_precision": 0.8619471744471745,
+      "eval_recall": 0.8896814075130766,
+      "eval_runtime": 44.15,
+      "eval_samples_per_second": 49.015,
+      "eval_steps_per_second": 6.138,
+      "step": 609
+    },
+    {
+      "epoch": 1.0673511293634497,
+      "grad_norm": 406.8649597167969,
+      "learning_rate": 8.371767994409504e-06,
+      "loss": 5.502258911132812,
+      "step": 650
+    },
+    {
+      "epoch": 1.1494866529774128,
+      "grad_norm": 353.2889709472656,
+      "learning_rate": 8.19706498951782e-06,
+      "loss": 5.161415405273438,
+      "step": 700
+    },
+    {
+      "epoch": 1.2316221765913757,
+      "grad_norm": 524.7266845703125,
+      "learning_rate": 8.022361984626136e-06,
+      "loss": 6.095115966796875,
+      "step": 750
+    },
+    {
+      "epoch": 1.3137577002053389,
+      "grad_norm": 532.918212890625,
+      "learning_rate": 7.847658979734451e-06,
+      "loss": 5.310472412109375,
+      "step": 800
+    },
+    {
+      "epoch": 1.3958932238193018,
+      "grad_norm": 481.7142639160156,
+      "learning_rate": 7.672955974842768e-06,
+      "loss": 5.16358154296875,
+      "step": 850
+    },
+    {
+      "epoch": 1.478028747433265,
+      "grad_norm": 427.3961181640625,
+      "learning_rate": 7.498252969951084e-06,
+      "loss": 5.792083740234375,
+      "step": 900
+    },
+    {
+      "epoch": 1.5601642710472279,
+      "grad_norm": 571.2908935546875,
+      "learning_rate": 7.3235499650594e-06,
+      "loss": 4.886295166015625,
+      "step": 950
+    },
+    {
+      "epoch": 1.642299794661191,
+      "grad_norm": 310.74542236328125,
+      "learning_rate": 7.148846960167716e-06,
+      "loss": 5.547396240234375,
+      "step": 1000
+    },
+    {
+      "epoch": 1.724435318275154,
+      "grad_norm": 210.18771362304688,
+      "learning_rate": 6.974143955276031e-06,
+      "loss": 5.537097778320312,
+      "step": 1050
+    },
+    {
+      "epoch": 1.806570841889117,
+      "grad_norm": 323.7012023925781,
+      "learning_rate": 6.799440950384347e-06,
+      "loss": 4.974610290527344,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8887063655030802,
+      "grad_norm": 300.61309814453125,
+      "learning_rate": 6.624737945492663e-06,
+      "loss": 5.379744262695312,
+      "step": 1150
+    },
+    {
+      "epoch": 1.9708418891170432,
+      "grad_norm": 420.67388916015625,
+      "learning_rate": 6.4500349406009785e-06,
+      "loss": 5.358375854492188,
+      "step": 1200
+    },
+    {
+      "epoch": 2.0,
+      "eval_f1": 0.8758669056339127,
+      "eval_loss": 4.113088607788086,
+      "eval_precision": 0.8614347026364194,
+      "eval_recall": 0.8907909335869393,
+      "eval_runtime": 44.1674,
+      "eval_samples_per_second": 48.995,
+      "eval_steps_per_second": 6.136,
+      "step": 1218
+    },
+    {
+      "epoch": 2.0525667351129364,
+      "grad_norm": 716.19775390625,
+      "learning_rate": 6.275331935709295e-06,
+      "loss": 5.002279052734375,
+      "step": 1250
+    },
+    {
+      "epoch": 2.1347022587268993,
+      "grad_norm": 269.1075439453125,
+      "learning_rate": 6.100628930817611e-06,
+      "loss": 4.724185180664063,
+      "step": 1300
+    },
+    {
+      "epoch": 2.2168377823408623,
+      "grad_norm": 386.7925720214844,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 4.566873779296875,
+      "step": 1350
+    },
+    {
+      "epoch": 2.2989733059548256,
+      "grad_norm": 348.7635803222656,
+      "learning_rate": 5.751222921034242e-06,
+      "loss": 4.5791323852539065,
+      "step": 1400
+    },
+    {
+      "epoch": 2.3811088295687886,
+      "grad_norm": 226.85601806640625,
+      "learning_rate": 5.576519916142558e-06,
+      "loss": 4.002293090820313,
+      "step": 1450
+    },
+    {
+      "epoch": 2.4632443531827515,
+      "grad_norm": 324.4167785644531,
+      "learning_rate": 5.401816911250875e-06,
+      "loss": 4.811328125,
+      "step": 1500
+    },
+    {
+      "epoch": 2.545379876796715,
+      "grad_norm": 265.19696044921875,
+      "learning_rate": 5.22711390635919e-06,
+      "loss": 4.3527978515625,
+      "step": 1550
+    },
+    {
+      "epoch": 2.6275154004106778,
+      "grad_norm": 255.17453002929688,
+      "learning_rate": 5.052410901467506e-06,
+      "loss": 4.515827026367187,
+      "step": 1600
+    },
+    {
+      "epoch": 2.7096509240246407,
+      "grad_norm": 343.71630859375,
+      "learning_rate": 4.877707896575822e-06,
+      "loss": 4.646475830078125,
+      "step": 1650
+    },
+    {
+      "epoch": 2.7917864476386036,
+      "grad_norm": 309.7668151855469,
+      "learning_rate": 4.703004891684137e-06,
+      "loss": 4.717439880371094,
+      "step": 1700
+    },
+    {
+      "epoch": 2.8739219712525665,
+      "grad_norm": 211.75811767578125,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 4.192425231933594,
+      "step": 1750
+    },
+    {
+      "epoch": 2.95605749486653,
+      "grad_norm": 471.5155334472656,
+      "learning_rate": 4.353598881900769e-06,
+      "loss": 5.051603088378906,
+      "step": 1800
+    },
+    {
+      "epoch": 3.0,
+      "eval_f1": 0.88046875,
+      "eval_loss": 4.196066379547119,
+      "eval_precision": 0.8681250962871668,
+      "eval_recall": 0.8931684894595023,
+      "eval_runtime": 43.3428,
+      "eval_samples_per_second": 49.928,
+      "eval_steps_per_second": 6.252,
+      "step": 1827
+    },
+    {
+      "epoch": 3.037782340862423,
+      "grad_norm": 235.75936889648438,
+      "learning_rate": 4.1788958770090845e-06,
+      "loss": 4.453041381835938,
+      "step": 1850
+    },
+    {
+      "epoch": 3.119917864476386,
+      "grad_norm": 343.1545715332031,
+      "learning_rate": 4.0041928721174006e-06,
+      "loss": 3.923203430175781,
+      "step": 1900
+    },
+    {
+      "epoch": 3.202053388090349,
+      "grad_norm": 272.7549743652344,
+      "learning_rate": 3.829489867225717e-06,
+      "loss": 4.473578796386719,
+      "step": 1950
+    },
+    {
+      "epoch": 3.284188911704312,
+      "grad_norm": 205.64874267578125,
+      "learning_rate": 3.6547868623340327e-06,
+      "loss": 3.830061340332031,
+      "step": 2000
+    },
+    {
+      "epoch": 3.3663244353182753,
+      "grad_norm": 508.5237121582031,
+      "learning_rate": 3.4800838574423483e-06,
+      "loss": 3.810235595703125,
+      "step": 2050
+    },
+    {
+      "epoch": 3.4484599589322382,
+      "grad_norm": 292.8867492675781,
+      "learning_rate": 3.3053808525506643e-06,
+      "loss": 3.886056213378906,
+      "step": 2100
+    },
+    {
+      "epoch": 3.530595482546201,
+      "grad_norm": 348.3034362792969,
+      "learning_rate": 3.13067784765898e-06,
+      "loss": 3.2952536010742186,
+      "step": 2150
+    },
+    {
+      "epoch": 3.612731006160164,
+      "grad_norm": 287.3987121582031,
+      "learning_rate": 2.9559748427672955e-06,
+      "loss": 4.358919067382812,
+      "step": 2200
+    },
+    {
+      "epoch": 3.6948665297741274,
+      "grad_norm": 186.44305419921875,
+      "learning_rate": 2.7812718378756116e-06,
+      "loss": 4.031048583984375,
+      "step": 2250
+    },
+    {
+      "epoch": 3.7770020533880904,
+      "grad_norm": 588.447509765625,
+      "learning_rate": 2.606568832983927e-06,
+      "loss": 4.055077819824219,
+      "step": 2300
+    },
+    {
+      "epoch": 3.8591375770020533,
+      "grad_norm": 305.3722229003906,
+      "learning_rate": 2.4318658280922432e-06,
+      "loss": 3.518572998046875,
+      "step": 2350
+    },
+    {
+      "epoch": 3.9412731006160167,
+      "grad_norm": 403.2993469238281,
+      "learning_rate": 2.2571628232005593e-06,
+      "loss": 4.462532958984375,
+      "step": 2400
+    },
+    {
+      "epoch": 4.0,
+      "eval_f1": 0.878969957081545,
+      "eval_loss": 4.369068145751953,
+      "eval_precision": 0.8656624654165386,
+      "eval_recall": 0.8926929782849897,
+      "eval_runtime": 42.2843,
+      "eval_samples_per_second": 51.177,
+      "eval_steps_per_second": 6.409,
+      "step": 2436
+    },
+    {
+      "epoch": 4.02299794661191,
+      "grad_norm": 281.2543029785156,
+      "learning_rate": 2.082459818308875e-06,
+      "loss": 3.9037939453125,
+      "step": 2450
+    },
+    {
+      "epoch": 4.105133470225873,
+      "grad_norm": 371.0896911621094,
+      "learning_rate": 1.907756813417191e-06,
+      "loss": 3.2074765014648436,
+      "step": 2500
+    },
+    {
+      "epoch": 4.187268993839836,
+      "grad_norm": 297.3424987792969,
+      "learning_rate": 1.7330538085255066e-06,
+      "loss": 3.682501220703125,
+      "step": 2550
+    },
+    {
+      "epoch": 4.269404517453799,
+      "grad_norm": 367.3387756347656,
+      "learning_rate": 1.5583508036338226e-06,
+      "loss": 3.2262432861328123,
+      "step": 2600
+    },
+    {
+      "epoch": 4.351540041067762,
+      "grad_norm": 250.43655395507812,
+      "learning_rate": 1.3836477987421384e-06,
+      "loss": 3.864920349121094,
+      "step": 2650
+    },
+    {
+      "epoch": 4.4336755646817245,
+      "grad_norm": 296.9447937011719,
+      "learning_rate": 1.2089447938504543e-06,
+      "loss": 3.5194046020507814,
+      "step": 2700
+    },
+    {
+      "epoch": 4.515811088295688,
+      "grad_norm": 321.5145263671875,
+      "learning_rate": 1.03424178895877e-06,
+      "loss": 3.8251571655273438,
+      "step": 2750
+    },
+    {
+      "epoch": 4.597946611909651,
+      "grad_norm": 273.0017395019531,
+      "learning_rate": 8.59538784067086e-07,
+      "loss": 3.709759521484375,
+      "step": 2800
+    },
+    {
+      "epoch": 4.680082135523614,
+      "grad_norm": 221.5639190673828,
+      "learning_rate": 6.84835779175402e-07,
+      "loss": 3.486819152832031,
+      "step": 2850
+    },
+    {
+      "epoch": 4.762217659137577,
+      "grad_norm": 214.02008056640625,
+      "learning_rate": 5.101327742837178e-07,
+      "loss": 3.5221002197265623,
+      "step": 2900
+    },
+    {
+      "epoch": 4.84435318275154,
+      "grad_norm": 300.3870544433594,
+      "learning_rate": 3.3542976939203357e-07,
+      "loss": 3.477908935546875,
+      "step": 2950
+    },
+    {
+      "epoch": 4.926488706365503,
+      "grad_norm": 416.9628601074219,
+      "learning_rate": 1.6072676450034943e-07,
+      "loss": 3.711861877441406,
+      "step": 3000
+    },
+    {
+      "epoch": 5.0,
+      "eval_f1": 0.8796317965519931,
+      "eval_loss": 4.447880744934082,
+      "eval_precision": 0.8660522273425499,
+      "eval_recall": 0.893644000634015,
+      "eval_runtime": 41.9192,
+      "eval_samples_per_second": 51.623,
+      "eval_steps_per_second": 6.465,
+      "step": 3045
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 3045,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 2,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 2
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3045/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97858c4638f99430a1879ac2e22898be13fdaf4eb03fabedc400f472205c63d0
+size 5201

checkpoint-609/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe0e30051ae237ef164c550f5dcb3e487285978129f5a1d1e04d6dd554fdb311
+size 498726232

checkpoint-609/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:742aae66fb774561fea8409d1a686f7e3904d50f03255cd0ac0334ed468c6a33
+size 992853771

checkpoint-609/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a49c3dda9a9b0c96cb6edbd8ac80d359fbc7c5de327759d73472394cb2862b1
+size 14645

checkpoint-609/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f92786f9857c13a7306e7bdedf19eec87a1017458a9774feeed8265292da2830
+size 1383

checkpoint-609/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3783a058d579a4bebde371fc21e863100872f2a65c8b89e47625047cbb941522
+size 1465

checkpoint-609/trainer_state.json

@@ -0,0 +1,138 @@
+{
+  "best_global_step": 609,
+  "best_metric": 0.8755947274003588,
+  "best_model_checkpoint": "./ner_crf_boundary_output/checkpoint-609",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 609,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08213552361396304,
+      "grad_norm": 289.9013977050781,
+      "learning_rate": 2.677595628415301e-06,
+      "loss": 7.03633056640625,
+      "step": 50
+    },
+    {
+      "epoch": 0.16427104722792607,
+      "grad_norm": 397.5795593261719,
+      "learning_rate": 5.409836065573772e-06,
+      "loss": 5.944567260742187,
+      "step": 100
+    },
+    {
+      "epoch": 0.2464065708418891,
+      "grad_norm": 1030.340576171875,
+      "learning_rate": 8.14207650273224e-06,
+      "loss": 6.386118774414062,
+      "step": 150
+    },
+    {
+      "epoch": 0.32854209445585214,
+      "grad_norm": 416.71173095703125,
+      "learning_rate": 9.944095038434662e-06,
+      "loss": 6.638558349609375,
+      "step": 200
+    },
+    {
+      "epoch": 0.4106776180698152,
+      "grad_norm": 379.8328857421875,
+      "learning_rate": 9.769392033542979e-06,
+      "loss": 5.887216186523437,
+      "step": 250
+    },
+    {
+      "epoch": 0.4928131416837782,
+      "grad_norm": 328.294921875,
+      "learning_rate": 9.594689028651294e-06,
+      "loss": 6.057736206054687,
+      "step": 300
+    },
+    {
+      "epoch": 0.5749486652977412,
+      "grad_norm": 306.49560546875,
+      "learning_rate": 9.41998602375961e-06,
+      "loss": 6.300731201171875,
+      "step": 350
+    },
+    {
+      "epoch": 0.6570841889117043,
+      "grad_norm": 406.1339111328125,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 5.523515625,
+      "step": 400
+    },
+    {
+      "epoch": 0.7392197125256673,
+      "grad_norm": 405.34136962890625,
+      "learning_rate": 9.070580013976241e-06,
+      "loss": 6.377322998046875,
+      "step": 450
+    },
+    {
+      "epoch": 0.8213552361396304,
+      "grad_norm": 219.61463928222656,
+      "learning_rate": 8.895877009084557e-06,
+      "loss": 5.774469604492188,
+      "step": 500
+    },
+    {
+      "epoch": 0.9034907597535934,
+      "grad_norm": 294.69683837890625,
+      "learning_rate": 8.721174004192873e-06,
+      "loss": 6.709805297851562,
+      "step": 550
+    },
+    {
+      "epoch": 0.9856262833675564,
+      "grad_norm": 397.64190673828125,
+      "learning_rate": 8.546470999301189e-06,
+      "loss": 5.980140991210938,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.8755947274003588,
+      "eval_loss": 4.1545538902282715,
+      "eval_precision": 0.8619471744471745,
+      "eval_recall": 0.8896814075130766,
+      "eval_runtime": 44.15,
+      "eval_samples_per_second": 49.015,
+      "eval_steps_per_second": 6.138,
+      "step": 609
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 3045,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 2,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "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": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-609/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97858c4638f99430a1879ac2e22898be13fdaf4eb03fabedc400f472205c63d0
+size 5201

config.json

@@ -0,0 +1,83 @@
+{
+  "base_model": "FacebookAI/roberta-base",
+  "num_labels": 35,
+  "label2id": {
+    "O": 0,
+    "B-PER": 1,
+    "I-PER": 2,
+    "B-ORG": 3,
+    "I-ORG": 4,
+    "B-LOC": 5,
+    "I-LOC": 6,
+    "B-FAC": 7,
+    "I-FAC": 8,
+    "B-DATE": 9,
+    "I-DATE": 10,
+    "B-TIME": 11,
+    "I-TIME": 12,
+    "B-EVT": 13,
+    "I-EVT": 14,
+    "B-PROD": 15,
+    "I-PROD": 16,
+    "B-MONEY": 17,
+    "I-MONEY": 18,
+    "B-PCT": 19,
+    "I-PCT": 20,
+    "B-LAW": 21,
+    "I-LAW": 22,
+    "B-NORP": 23,
+    "I-NORP": 24,
+    "B-LANG": 25,
+    "I-LANG": 26,
+    "B-WOA": 27,
+    "I-WOA": 28,
+    "B-QTY": 29,
+    "I-QTY": 30,
+    "B-ORD": 31,
+    "I-ORD": 32,
+    "B-CARD": 33,
+    "I-CARD": 34
+  },
+  "id2label": {
+    "0": "O",
+    "1": "B-PER",
+    "2": "I-PER",
+    "3": "B-ORG",
+    "4": "I-ORG",
+    "5": "B-LOC",
+    "6": "I-LOC",
+    "7": "B-FAC",
+    "8": "I-FAC",
+    "9": "B-DATE",
+    "10": "I-DATE",
+    "11": "B-TIME",
+    "12": "I-TIME",
+    "13": "B-EVT",
+    "14": "I-EVT",
+    "15": "B-PROD",
+    "16": "I-PROD",
+    "17": "B-MONEY",
+    "18": "I-MONEY",
+    "19": "B-PCT",
+    "20": "I-PCT",
+    "21": "B-LAW",
+    "22": "I-LAW",
+    "23": "B-NORP",
+    "24": "I-NORP",
+    "25": "B-LANG",
+    "26": "I-LANG",
+    "27": "B-WOA",
+    "28": "I-WOA",
+    "29": "B-QTY",
+    "30": "I-QTY",
+    "31": "B-ORD",
+    "32": "I-ORD",
+    "33": "B-CARD",
+    "34": "I-CARD"
+  },
+  "boundary_loss_weight": 0.3,
+  "boundary_smoothing_eps": 0.1,
+  "fgm_epsilon": 1.0,
+  "architecture": "RobertaCRFBoundaryForNER",
+  "notes": "RoBERTa encoder + linear classifier + CRF, extended with a boundary-guided interaction module (boundary_start_head/boundary_end_head), boundary-smoothed multi-task loss, and FGM adversarial training. Fine-tuned starting from Souvikbasur/roberta-crf-news-ner."
+}

confusion_matrix_boundary.csv

@@ -0,0 +1,19 @@
+gold\pred,CARD,DATE,EVT,FAC,LANG,LAW,LOC,MISSED,MONEY,NORP,ORD,ORG,PCT,PER,PROD,QTY,SPURIOUS,TIME,WOA
+CARD,441,1,0,0,0,0,0,34,0,0,0,0,0,0,0,2,0,0,0
+DATE,1,810,0,0,0,0,0,103,0,0,0,0,0,0,0,0,0,4,0
+EVT,0,2,69,0,0,0,0,19,0,0,0,1,0,2,0,0,0,0,0
+FAC,0,0,0,40,0,0,3,12,0,0,0,9,0,0,0,0,0,0,0
+LANG,0,0,0,0,24,0,0,2,0,0,0,0,0,0,0,0,0,0,0
+LAW,0,0,0,0,0,50,0,17,0,0,0,1,0,0,0,0,0,0,0
+LOC,0,0,0,6,0,0,1232,60,0,2,0,10,0,3,1,0,0,1,1
+MONEY,0,0,0,0,0,0,0,21,200,0,0,0,0,0,1,0,0,0,0
+NORP,0,0,0,0,3,0,1,30,0,501,0,6,0,2,0,0,0,0,1
+ORD,0,1,0,0,0,0,0,4,0,0,109,0,0,0,0,0,0,0,0
+ORG,0,0,0,5,0,4,4,179,0,5,0,1199,0,14,6,0,0,0,9
+PCT,1,0,0,0,0,0,0,12,0,0,0,0,139,0,0,0,0,0,0
+PER,0,0,0,0,0,2,5,40,0,1,0,6,0,1056,1,0,0,0,1
+PROD,0,0,0,1,0,0,0,16,0,0,0,10,0,1,122,0,0,0,1
+QTY,7,4,0,0,0,0,0,21,0,0,0,0,0,0,0,69,0,0,0
+SPURIOUS,70,129,17,14,1,25,59,0,22,41,20,215,18,61,19,23,0,18,49
+TIME,0,3,0,0,0,0,0,12,0,0,0,0,0,0,0,0,0,64,0
+WOA,0,1,1,0,0,1,0,34,0,0,0,5,0,0,3,0,0,0,91

final_test_report_crf_boundary.txt

@@ -0,0 +1,23 @@
+              precision    recall  f1-score   support
+
+        CARD     0.8464    0.9245    0.8838       477
+        DATE     0.8538    0.8982    0.8755       904
+         EVT     0.8161    0.7634    0.7889        93
+         FAC     0.6061    0.6250    0.6154        64
+        LANG     0.8571    0.9231    0.8889        26
+         LAW     0.6463    0.7794    0.7067        68
+         LOC     0.9471    0.9413    0.9442      1312
+       MONEY     0.9009    0.9009    0.9009       222
+        NORP     0.9309    0.9412    0.9360       544
+         ORD     0.8450    0.9561    0.8971       114
+         ORG     0.8323    0.8539    0.8430      1424
+         PCT     0.8854    0.9145    0.8997       152
+         PER     0.9350    0.9612    0.9479      1108
+        PROD     0.8170    0.8278    0.8224       151
+         QTY     0.7340    0.6832    0.7077       101
+        TIME     0.7356    0.8101    0.7711        79
+         WOA     0.6013    0.6765    0.6367       136
+
+   micro avg     0.8724    0.8985    0.8852      6975
+   macro avg     0.8112    0.8459    0.8274      6975
+weighted avg     0.8738    0.8985    0.8857      6975

roberta_crf_boundary_ner.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59da642e67f12de5bb7cd11d1fb3f803f90832df902a9c8876d260f15a08721f
+size 498790741

tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "cls_token": "<s>",
+  "eos_token": "</s>",
+  "errors": "replace",
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 512,
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "tokenizer_class": "RobertaTokenizer",
+  "trim_offsets": true,
+  "unk_token": "<unk>"
+}