Model save

README.md

@@ -1,27 +1,27 @@
 ---
-license: apache-2.0
-base_model: google-bert/bert-base-multilingual-cased
+license: mit
+base_model: microsoft/mdeberta-v3-base
 tags:
 - generated_from_trainer
 metrics:
 - precision
 - recall
 model-index:
-- name: lex-cross-encoder-mbert-10neg
+- name: lex-cross-encoder-mdeberta-v3-base-5neg
   results: []
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# lex-cross-encoder-mbert-10neg
+# lex-cross-encoder-mdeberta-v3-base-5neg
 
-This model is a fine-tuned version of [google-bert/bert-base-multilingual-cased](https://huggingface.co/google-bert/bert-base-multilingual-cased) on an unknown dataset.
+This model is a fine-tuned version of [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4360
-- Precision: 0.6020
-- Recall: 0.8593
-- F2: 0.7917
+- Loss: 0.6811
+- Precision: 0.2
+- Recall: 1.0
+- F2: 0.5556
 
 ## Model description
 
@@ -40,38 +40,30 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 1e-05
-- train_batch_size: 16
-- eval_batch_size: 16
+- learning_rate: 1e-06
+- train_batch_size: 8
+- eval_batch_size: 8
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 8
-- total_train_batch_size: 128
-- total_eval_batch_size: 128
+- total_train_batch_size: 64
+- total_eval_batch_size: 64
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 10
+- num_epochs: 2
 
 ### Training results
 
-| Training Loss | Epoch | Step  | Validation Loss | Precision | Recall | F2     |
-|:-------------:|:-----:|:-----:|:---------------:|:---------:|:------:|:------:|
-| 0.4572        | 1.0   | 2317  | 0.4705          | 0.4735    | 0.8620 | 0.7405 |
-| 0.4283        | 2.0   | 4634  | 0.4515          | 0.4774    | 0.9124 | 0.7718 |
-| 0.4115        | 3.0   | 6951  | 0.4485          | 0.4796    | 0.9201 | 0.7773 |
-| 0.4021        | 4.0   | 9268  | 0.4387          | 0.5217    | 0.9068 | 0.7902 |
-| 0.3918        | 5.0   | 11585 | 0.4466          | 0.6111    | 0.8242 | 0.7705 |
-| 0.3879        | 6.0   | 13902 | 0.4337          | 0.5783    | 0.8767 | 0.7947 |
-| 0.383         | 7.0   | 16219 | 0.4336          | 0.5633    | 0.8907 | 0.7980 |
-| 0.3781        | 8.0   | 18536 | 0.4354          | 0.5929    | 0.8660 | 0.7930 |
-| 0.3767        | 9.0   | 20853 | 0.4353          | 0.5980    | 0.8636 | 0.7931 |
-| 0.3712        | 10.0  | 23170 | 0.4360          | 0.6020    | 0.8593 | 0.7917 |
+| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F2     |
+|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|
+| 0.7455        | 1.0   | 1    | 0.6810          | 0.2       | 1.0    | 0.5556 |
+| 0.7455        | 2.0   | 2    | 0.6811          | 0.2       | 1.0    | 0.5556 |
 
 
 ### Framework versions
 
 - Transformers 4.39.1
-- Pytorch 2.5.1+cu121
+- Pytorch 2.6.0+cu124
 - Datasets 3.6.0
 - Tokenizers 0.15.2

final_model/config.json

@@ -1,50 +1,35 @@
 {
-  "_name_or_path": "Alibaba-NLP/gte-multilingual-reranker-base",
+  "_name_or_path": "microsoft/mdeberta-v3-base",
   "architectures": [
-    "NewForSequenceClassification"
+    "DebertaV2ForSequenceClassification"
   ],
-  "attention_probs_dropout_prob": 0.0,
-  "auto_map": {
-    "AutoConfig": "Alibaba-NLP/new-impl--configuration.NewConfig",
-    "AutoModel": "Alibaba-NLP/new-impl--modeling.NewModel",
-    "AutoModelForMaskedLM": "Alibaba-NLP/new-impl--modeling.NewForMaskedLM",
-    "AutoModelForMultipleChoice": "Alibaba-NLP/new-impl--modeling.NewForMultipleChoice",
-    "AutoModelForQuestionAnswering": "Alibaba-NLP/new-impl--modeling.NewForQuestionAnswering",
-    "AutoModelForSequenceClassification": "Alibaba-NLP/new-impl--modeling.NewForSequenceClassification",
-    "AutoModelForTokenClassification": "Alibaba-NLP/new-impl--modeling.NewForTokenClassification"
-  },
-  "classifier_dropout": 0.0,
+  "attention_probs_dropout_prob": 0.1,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
-  "id2label": {
-    "0": "LABEL_0"
-  },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
-  "label2id": {
-    "LABEL_0": 0
-  },
-  "layer_norm_eps": 1e-12,
-  "layer_norm_type": "layer_norm",
-  "logn_attention_clip1": false,
-  "logn_attention_scale": false,
-  "max_position_embeddings": 8192,
-  "model_type": "new",
+  "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,
-  "pack_qkv": true,
-  "pad_token_id": 1,
-  "position_embedding_type": "rope",
-  "rope_scaling": {
-    "factor": 8.0,
-    "type": "ntk"
-  },
-  "rope_theta": 20000,
+  "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": "4.39.1",
-  "type_vocab_size": 1,
-  "unpad_inputs": false,
-  "use_memory_efficient_attention": false,
-  "vocab_size": 250048
+  "type_vocab_size": 0,
+  "vocab_size": 251000
 }

final_model/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0d096a2bf9e6d3723ca116fc08bcd30246856f821d5f5723dcbc067101e4b20c
-size 1223854204
+oid sha256:89c03989f75d0aa4bd09aa494db2abf9fdb0699814577f09a2b8630feeff3136
+size 1115268200

final_model/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:faf4985ae8e249f909a317e50716aa5e88a5b360258c73b674d86bb7fe61a012
-size 5176
+oid sha256:8c35a0a052756eed7c3a8019317fe0fdeda73d19c1f4811fdf8efcef22ec1a7d
+size 5048