Upload folder using huggingface_hub

config.json

@@ -6,7 +6,59 @@
     "AutoConfig": "configuration_comparer.ComparerConfig",
     "AutoModel": "modeling_comparer.ComparerModel"
   },
-  "base_model": "microsoft/deberta-v3-base",
+  "base_config_dict": {
+    "_name_or_path": "AI-Response-Comparer/deberta-v3-base",
+    "architectures": [
+      "DebertaV2Model"
+    ],
+    "attention_probs_dropout_prob": 0.1,
+    "bos_token_id": null,
+    "chunk_size_feed_forward": 0,
+    "dtype": "float16",
+    "eos_token_id": null,
+    "hidden_act": "gelu",
+    "hidden_dropout_prob": 0.1,
+    "hidden_size": 768,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "intermediate_size": 3072,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "layer_norm_eps": 1e-07,
+    "legacy": true,
+    "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,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "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,
+    "problem_type": null,
+    "relative_attention": true,
+    "return_dict": true,
+    "share_att_key": true,
+    "tie_word_embeddings": true,
+    "transformers_version": "5.2.0",
+    "type_vocab_size": 0,
+    "vocab_size": 128100
+  },
   "dtype": "float16",
   "hidden_size": 768,
   "model_type": "comparer",

configuration_comparer.py

@@ -1,21 +1,9 @@
-from transformers import PretrainedConfig, AutoConfig
+from transformers import PretrainedConfig
 
 class ComparerConfig(PretrainedConfig):
     model_type = "comparer"
 
-    def __init__(
-        self,
-        base_model="microsoft/deberta-v3-base",
-        hidden_size=768,
-        base_model_config=None,
-        **kwargs
-    ):
+    def __init__(self, hidden_size=768, base_config_dict=None, **kwargs):
         super().__init__(**kwargs)
-
-        self.base_model = base_model
         self.hidden_size = hidden_size
-
-        if base_model_config is None:
-            base_model_config = AutoConfig.from_pretrained(base_model)
-
-        self.base_model_config = base_model_config
+        self.base_config_dict = base_config_dict or {}

modeling_comparer.py

@@ -1,7 +1,7 @@
 import torch
 import torch.nn as nn
-from transformers import PreTrainedModel, AutoModel
-from .configuration_comparer import ComparerConfig
+from transformers import PreTrainedModel, AutoModel, AutoConfig
+from configuration_comparer import ComparerConfig
 
 class ComparerModel(PreTrainedModel):
     config_class = ComparerConfig
@@ -9,7 +9,16 @@ class ComparerModel(PreTrainedModel):
     def __init__(self, config):
         super().__init__(config)
 
-        self.main_layer = AutoModel.from_config(config.base_model_config)
+        if not config.base_config_dict:
+            raise ValueError("base_config_dict is missing from the configuration! Cannot build base model.")
+
+        base_dict = config.base_config_dict.copy()
+        
+        model_type = base_dict.pop("model_type", "deberta-v2")
+        
+        base_config = AutoConfig.for_model(model_type, **base_dict)
+        
+        self.main_layer = AutoModel.from_config(base_config)
 
         self.linear_tokens = nn.Linear(config.hidden_size, 100)
         self.linear_tokens2 = nn.Linear(200, 50)

tokenizer_config.json

@@ -10,7 +10,7 @@
     "[CLS]",
     "[SEP]"
   ],
-  "is_local": false,
+  "is_local": true,
   "mask_token": "[MASK]",
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "[PAD]",