Fix model name from InternVL2-8B to InternVL2

app.py

@@ -104,7 +104,7 @@ def create_in_memory_flash_attn_mock():
     # Add attributes used by transformers checks
     flash_attn.__version__ = "0.0.0-mocked"
     
-    # Create common submodules - FIXED to not use 'parent' parameter in ModuleSpec
+    # Create common submodules - without 'parent' parameter
     for submodule in ['flash_attn.flash_attn_interface', 'flash_attn.flash_attn_triton']:
         parts = submodule.split('.')
         parent_name = '.'.join(parts[:-1])
@@ -170,7 +170,8 @@ def load_model():
         print("\nLoading InternVL2 model...")
         
         # Load the model and tokenizer
-        model_path = "OpenGVLab/InternVL2-8B"
+        # FIXED: Corrected model name from InternVL2-8B to InternVL2
+        model_path = "OpenGVLab/InternVL2"
         
         # Print downloading status
         print("Downloading model shards. This may take some time...")