Upload modeling_conceptframemet.py with huggingface_hub

modeling_conceptframemet.py

@@ -55,6 +55,18 @@ class AdaptiveSourceQAMelBert(nn.Module):
         super(AdaptiveSourceQAMelBert, self).__init__()
         self.num_labels = num_labels
         self.encoder = Model
+        
+        # FIX: Resize token_type_embeddings to match training (type_vocab_size=4)
+        if hasattr(self.encoder, 'embeddings') and hasattr(self.encoder.embeddings, 'token_type_embeddings'):
+            if self.encoder.embeddings.token_type_embeddings.weight.shape[0] != 4:
+                old_embeddings = self.encoder.embeddings.token_type_embeddings
+                new_embeddings = nn.Embedding(4, old_embeddings.embedding_dim)
+                new_embeddings.weight.data[0] = old_embeddings.weight.data[0]
+                new_embeddings.weight.data[1:].normal_(mean=0.0, std=config.initializer_range)
+                self.encoder.embeddings.token_type_embeddings = new_embeddings
+                if hasattr(self.encoder, 'config'):
+                    self.encoder.config.type_vocab_size = 4
+        
         self.source_qa_model = Source_QA_Model
         self.source_qa_tokenizer = source_qa_tokenizer
         self.melbert_tokenizer = melbert_tokenizer
@@ -81,10 +93,26 @@ class AdaptiveSourceQAMelBert(nn.Module):
         self.source_id2label = {}
         try:
             import json
-            with open('source_finder/source_labels.json', 'r') as f:
-                source_label2id = json.load(f)
-                self.source_id2label = {v: k for k, v in source_label2id.items()}
-                print(f"✓ Loaded {len(self.source_id2label)} source domain labels")
+            import os
+            # Try multiple paths
+            possible_paths = [
+                'source_labels.json',  # Same directory as model file
+                'source_finder/source_labels.json',  # Original location
+                os.path.join(os.path.dirname(__file__), 'source_labels.json'),  # Next to this file
+            ]
+            
+            for path in possible_paths:
+                try:
+                    with open(path, 'r') as f:
+                        source_label2id = json.load(f)
+                        self.source_id2label = {v: k for k, v in source_label2id.items()}
+                        print(f"✓ Loaded {len(self.source_id2label)} source domain labels from {path}")
+                        break
+                except:
+                    continue
+            
+            if not self.source_id2label:
+                print(f"❌ Warning: Could not load source labels from any location")
         except Exception as e:
             print(f"❌ Warning: Could not load source labels: {e}")