Upload app.py with huggingface_hub

app.py

@@ -37,49 +37,27 @@ from src.utils.config import load_config, merge_configs
 model = None
 config = None
 
-def download_and_load_model():
-    """Download and load the trained multimodal model from HF"""
+def load_model():
+    """Load the trained multimodal model from local files"""
     global model, config
 
     if model is not None:
         return "✅ Model already loaded!"
 
     try:
-        print("🔄 Downloading multimodal Gemma model from HF...")
-
-        # Download model checkpoint (using token if available)
-        checkpoint_path = hf_hub_download(
-            repo_id="sagar007/multimodal-gemma-270m-llava",
-            filename="final_model.ckpt",
-            cache_dir="./model_cache",
-            token=hf_token if 'hf_token' in locals() else None
-        )
+        print("🔄 Loading multimodal Gemma model from local files...")
 
-        # Download config files (same as local setup)
-        model_config_path = hf_hub_download(
-            repo_id="sagar007/multimodal-gemma-270m-llava",
-            filename="configs/model_config.yaml",
-            cache_dir="./model_cache",
-            token=hf_token if 'hf_token' in locals() else None
-        )
-        training_config_path = hf_hub_download(
-            repo_id="sagar007/multimodal-gemma-270m-llava",
-            filename="configs/training_config.yaml",
-            cache_dir="./model_cache",
-            token=hf_token if 'hf_token' in locals() else None
-        )
-        data_config_path = hf_hub_download(
-            repo_id="sagar007/multimodal-gemma-270m-llava",
-            filename="configs/data_config.yaml",
-            cache_dir="./model_cache",
-            token=hf_token if 'hf_token' in locals() else None
-        )
+        # Use local model checkpoint (included in Space)
+        checkpoint_path = "final_model.ckpt"
+
+        if not Path(checkpoint_path).exists():
+            return "❌ Model checkpoint not found! Please ensure final_model.ckpt is in the Space."
 
-        # Load configs exactly like local gradio_app.py
+        # Use local config files (same as local setup)
         print("📁 Loading configs...")
-        model_config = load_config(model_config_path)
-        training_config = load_config(training_config_path)
-        data_config = load_config(data_config_path)
+        model_config = load_config("configs/model_config.yaml")
+        training_config = load_config("configs/training_config.yaml")
+        data_config = load_config("configs/data_config.yaml")
         config = merge_configs([model_config, training_config, data_config])
 
         print("📁 Loading model from checkpoint...")
@@ -334,7 +312,7 @@ def create_gradio_interface():
 
         # Event handlers
         load_btn.click(
-            fn=download_and_load_model,
+            fn=load_model,
             outputs=model_status
         )