Initial Commit

README.md

@@ -1,6 +1,6 @@
 ---
 license: mit
-base_model: facebook/xlm-v-base
+base_model: microsoft/mdeberta-v3-base
 tags:
 - generated_from_trainer
 datasets:
@@ -23,10 +23,10 @@ model-index:
     metrics:
     - name: Accuracy
       type: accuracy
-      value: 0.051647811116576486
+      value: 0.8558780127889818
     - name: F1
       type: f1
-      value: 0.0016647904742274576
+      value: 0.8318635435156069
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -34,11 +34,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 # scenario-TCR_data-AmazonScience_massive_all_1_1
 
-This model is a fine-tuned version of [facebook/xlm-v-base](https://huggingface.co/facebook/xlm-v-base) on the massive dataset.
+This model is a fine-tuned version of [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base) on the massive dataset.
 It achieves the following results on the evaluation set:
-- Loss: 3.8189
-- Accuracy: 0.0516
-- F1: 0.0017
+- Loss: 0.9483
+- Accuracy: 0.8559
+- F1: 0.8319
 
 ## Model description
 
@@ -60,22 +60,33 @@ The following hyperparameters were used during training:
 - learning_rate: 5e-05
 - train_batch_size: 32
 - eval_batch_size: 32
-- seed: 42
+- seed: 66
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 500
+- num_epochs: 30
 
 ### Training results
 
 | Training Loss | Epoch | Step  | Validation Loss | Accuracy | F1     |
 |:-------------:|:-----:|:-----:|:---------------:|:--------:|:------:|
-| 3.7541        | 0.27  | 5000  | 3.7437          | 0.0605   | 0.0019 |
-| 3.7191        | 0.53  | 10000 | 3.7655          | 0.0620   | 0.0020 |
-| 3.7281        | 0.8   | 15000 | 3.8011          | 0.0516   | 0.0017 |
-| 3.7116        | 1.07  | 20000 | 3.8592          | 0.0516   | 0.0017 |
-| 3.7068        | 1.34  | 25000 | 3.8168          | 0.0516   | 0.0017 |
-| 3.7064        | 1.6   | 30000 | 3.8746          | 0.0516   | 0.0017 |
-| 3.7059        | 1.87  | 35000 | 3.8189          | 0.0516   | 0.0017 |
+| 0.519         | 0.27  | 5000  | 0.6915          | 0.8379   | 0.7941 |
+| 0.3806        | 0.53  | 10000 | 0.6969          | 0.8468   | 0.8063 |
+| 0.3259        | 0.8   | 15000 | 0.6916          | 0.8515   | 0.8159 |
+| 0.2379        | 1.07  | 20000 | 0.7826          | 0.8505   | 0.8191 |
+| 0.236         | 1.34  | 25000 | 0.7514          | 0.8508   | 0.8189 |
+| 0.2298        | 1.6   | 30000 | 0.7719          | 0.8526   | 0.8267 |
+| 0.2169        | 1.87  | 35000 | 0.8162          | 0.8505   | 0.8265 |
+| 0.164         | 2.14  | 40000 | 0.8316          | 0.8549   | 0.8272 |
+| 0.1684        | 2.41  | 45000 | 0.8123          | 0.8513   | 0.8204 |
+| 0.158         | 2.67  | 50000 | 0.8252          | 0.8556   | 0.8309 |
+| 0.1761        | 2.94  | 55000 | 0.8092          | 0.8545   | 0.8287 |
+| 0.1378        | 3.21  | 60000 | 0.8574          | 0.8607   | 0.8357 |
+| 0.1399        | 3.47  | 65000 | 0.8976          | 0.8572   | 0.8359 |
+| 0.1431        | 3.74  | 70000 | 0.8908          | 0.8536   | 0.8350 |
+| 0.1249        | 4.01  | 75000 | 0.9613          | 0.8533   | 0.8292 |
+| 0.1129        | 4.28  | 80000 | 0.9511          | 0.8543   | 0.8306 |
+| 0.1143        | 4.54  | 85000 | 0.9001          | 0.8563   | 0.8331 |
+| 0.122         | 4.81  | 90000 | 0.9483          | 0.8559   | 0.8319 |
 
 
 ### Framework versions

config.json

@@ -1,12 +1,9 @@
 {
-  "_name_or_path": "facebook/xlm-v-base",
+  "_name_or_path": "microsoft/mdeberta-v3-base",
   "architectures": [
-    "XLMRobertaForSequenceClassification"
+    "DebertaV2ForSequenceClassification"
   ],
   "attention_probs_dropout_prob": 0.1,
-  "bos_token_id": 0,
-  "classifier_dropout": null,
-  "eos_token_id": 2,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
@@ -136,17 +133,27 @@
     "LABEL_8": 8,
     "LABEL_9": 9
   },
-  "layer_norm_eps": 1e-05,
-  "max_position_embeddings": 514,
-  "model_type": "xlm-roberta",
+  "layer_norm_eps": 1e-07,
+  "max_position_embeddings": 512,
+  "max_relative_positions": -1,
+  "model_type": "deberta-v2",
+  "norm_rel_ebd": "layer_norm",
   "num_attention_heads": 12,
   "num_hidden_layers": 12,
-  "pad_token_id": 1,
-  "position_embedding_type": "absolute",
-  "problem_type": "single_label_classification",
+  "pad_token_id": 0,
+  "pooler_dropout": 0,
+  "pooler_hidden_act": "gelu",
+  "pooler_hidden_size": 768,
+  "pos_att_type": [
+    "p2c",
+    "c2p"
+  ],
+  "position_biased_input": false,
+  "position_buckets": 256,
+  "relative_attention": true,
+  "share_att_key": true,
   "torch_dtype": "float32",
   "transformers_version": "4.33.3",
-  "type_vocab_size": 1,
-  "use_cache": true,
-  "vocab_size": 901629
+  "type_vocab_size": 0,
+  "vocab_size": 251000
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f71b68528e7dfc888fd501d031df288e55220a595fedd44e0a074c4489ea6c6
-size 3114226670
+oid sha256:46e4599edc4a5f0d569a174684dbef4435918fc6cb26d0986eb6eab29335e22a
+size 1115491954

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cfd5c5598e502bf60285c2298617fec9b1aa82429d2620bf1eaf88144f6746de
+oid sha256:5415336326de347ef5dd173e77a1892eb5faece5e0a9bdf316852ba765f25a5a
 size 4600