upload models/dataset-sciarg/task-adur/cluster/2025-03-12_02-12-01

model is from: https://github.com/ArneBinder/pie-document-level/pull/410#issuecomment-2711611125
- evaluated on SciArg validation set (from sciarg_prepared_with_fixed_validation_and_more_train)
- selection based on relaxed_span=true and criterion=span: idx 4

config.json

@@ -1,47 +1,47 @@
 {
-  "model_type": "TokenClassificationModelWithSeq2SeqEncoderAndCrf",
-  "model_name_or_path": "allenai/scibert_scivocab_uncased",
-  "num_classes": 7,
-  "learning_rate": 5e-05,
-  "task_learning_rate": 0.005,
-  "use_crf": true,
-  "label_pad_id": -100,
-  "special_token_label_id": 0,
-  "classifier_dropout": null,
-  "freeze_base_model": false,
-  "warmup_proportion": 0.1,
-  "seq2seq_encoder": {
-    "type": "sequential",
-    "drop0": {
-      "type": "dropout",
-      "p": 0.5
-    },
-    "lstm0": {
-      "type": "lstm",
-      "num_layers": 2,
-      "bidirectional": true,
-      "hidden_size": 300,
-      "dropout": 0.4394
-    }
-  },
-  "taskmodule_config": {
-    "taskmodule_type": "LabeledSpanExtractionByTokenClassificationTaskModule",
-    "tokenizer_name_or_path": "allenai/scibert_scivocab_uncased",
-    "span_annotation": "labeled_spans",
-    "partition_annotation": "labeled_partitions",
+    "model_type": "TokenClassificationModelWithSeq2SeqEncoderAndCrf",
+    "model_name_or_path": "allenai/scibert_scivocab_uncased",
+    "num_classes": 7,
+    "learning_rate": 5e-05,
+    "task_learning_rate": 0.01,
+    "use_crf": true,
     "label_pad_id": -100,
-    "labels": [
-      "background_claim",
-      "data",
-      "own_claim"
-    ],
-    "include_ill_formed_predictions": true,
-    "tokenize_kwargs": {
-      "max_length": 512,
-      "stride": 64,
-      "return_overflowing_tokens": true
+    "special_token_label_id": 0,
+    "classifier_dropout": null,
+    "freeze_base_model": true,
+    "warmup_proportion": 0.1,
+    "seq2seq_encoder": {
+        "type": "sequential",
+        "drop0": {
+            "type": "dropout",
+            "p": 0.5
+        },
+        "lstm0": {
+            "type": "lstm",
+            "num_layers": 2,
+            "bidirectional": true,
+            "hidden_size": 300,
+            "dropout": 0.4394
+        }
     },
-    "pad_kwargs": null,
-    "log_precision_recall_metrics": true
-  }
+    "taskmodule_config": {
+        "taskmodule_type": "LabeledSpanExtractionByTokenClassificationTaskModule",
+        "tokenizer_name_or_path": "allenai/scibert_scivocab_uncased",
+        "span_annotation": "labeled_spans",
+        "partition_annotation": "labeled_partitions",
+        "label_pad_id": -100,
+        "labels": [
+            "background_claim",
+            "data",
+            "own_claim"
+        ],
+        "include_ill_formed_predictions": true,
+        "tokenize_kwargs": {
+            "max_length": 512,
+            "stride": 64,
+            "return_overflowing_tokens": true
+        },
+        "pad_kwargs": null,
+        "log_precision_recall_metrics": true
+    }
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a183e009828e2fd988f97e52b5fd5a4f1eca39b65abc97b35b99498f7b0d9930
-size 458710913
+oid sha256:89133bb2f468981542946e3f5b19d3b80c3dd326e35a97a2a865ceba2c930297
+size 458711358

taskmodule_config.json

@@ -1,20 +1,20 @@
 {
-  "taskmodule_type": "LabeledSpanExtractionByTokenClassificationTaskModule",
-  "tokenizer_name_or_path": "allenai/scibert_scivocab_uncased",
-  "span_annotation": "labeled_spans",
-  "partition_annotation": "labeled_partitions",
-  "label_pad_id": -100,
-  "labels": [
-    "background_claim",
-    "data",
-    "own_claim"
-  ],
-  "include_ill_formed_predictions": true,
-  "tokenize_kwargs": {
-    "max_length": 512,
-    "stride": 64,
-    "return_overflowing_tokens": true
-  },
-  "pad_kwargs": null,
-  "log_precision_recall_metrics": true
+    "taskmodule_type": "LabeledSpanExtractionByTokenClassificationTaskModule",
+    "tokenizer_name_or_path": "allenai/scibert_scivocab_uncased",
+    "span_annotation": "labeled_spans",
+    "partition_annotation": "labeled_partitions",
+    "label_pad_id": -100,
+    "labels": [
+        "background_claim",
+        "data",
+        "own_claim"
+    ],
+    "include_ill_formed_predictions": true,
+    "tokenize_kwargs": {
+        "max_length": 512,
+        "stride": 64,
+        "return_overflowing_tokens": true
+    },
+    "pad_kwargs": null,
+    "log_precision_recall_metrics": true
 }