Update app.py

app.py

@@ -21,10 +21,10 @@ if not HF_TOKEN:
     raise ValueError("HUGGINGFACE_TOKEN environment variable is not set")
 
 # Hardcoded model repository ID
-HARDCODED_MODEL_REPO_ID = "openai/gpt-oss-20b"  # Change this to your desired model
+HARDCODED_MODEL_REPO_ID = "openai/gpt-oss-20b"  # Your model
 
 # Hardcoded dataset repository ID
-HARDCODED_DATASET_REPO_ID = "Fred808/helium_memory"  # Change this to your dataset
+HARDCODED_DATASET_REPO_ID = "Fred808/helium_memory"  # Your dataset
 
 # Hardcoded path in repository
 HARDCODED_PATH_IN_REPO = "model_data/"
@@ -52,7 +52,7 @@ def download_model_files() -> str:
         
         print(f"Found {len(files)} files to download")
         
-        # Download each file
+        # Download each file directly without using cache
         for file_path in files:
             if file_path.endswith('/'):  # Skip directories
                 continue
@@ -63,22 +63,26 @@ def download_model_files() -> str:
             file_dir = os.path.join(temp_dir, os.path.dirname(file_path))
             os.makedirs(file_dir, exist_ok=True)
             
-            # Download the file
+            # Download the file directly to the target location
             local_path = hf_hub_download(
                 repo_id=HARDCODED_MODEL_REPO_ID,
                 filename=file_path,
-                cache_dir=temp_dir,
+                local_dir=temp_dir,
                 force_download=True,
-                resume_download=False,
                 token=HF_TOKEN
             )
             
-            # Move from cache to the desired location
-            final_path = os.path.join(temp_dir, file_path)
-            if local_path != final_path:
-                shutil.move(local_path, final_path)
+            print(f"Downloaded to: {local_path}")
+        
+        # Remove any cache directories that might have been created
+        for item in os.listdir(temp_dir):
+            item_path = os.path.join(temp_dir, item)
+            if item.startswith('models--') and os.path.isdir(item_path):
+                shutil.rmtree(item_path)
+                print(f"Removed cache directory: {item_path}")
         
         print(f"All files downloaded to: {temp_dir}")
+        print(f"Final directory contents: {os.listdir(temp_dir)}")
         return temp_dir
         
     except Exception as e:
@@ -91,14 +95,28 @@ def upload_folder_to_dataset(folder_path: str):
     """Uploads a folder to the hardcoded Hugging Face dataset repository."""
     api = HfApi(token=HF_TOKEN)
     print(f"Uploading {folder_path} to {HARDCODED_DATASET_REPO_ID} at {HARDCODED_PATH_IN_REPO}...")
+    
+    # Check what we're actually trying to upload
+    print(f"Folder contents: {os.listdir(folder_path)}")
+    
     try:
-        api.upload_folder(
-            folder_path=folder_path,
-            path_in_repo=HARDCODED_PATH_IN_REPO,
-            repo_id=HARDCODED_DATASET_REPO_ID,
-            repo_type="dataset",
-            token=HF_TOKEN
-        )
+        # Upload each file individually to avoid cache issues
+        for root, dirs, files in os.walk(folder_path):
+            for file in files:
+                file_path = os.path.join(root, file)
+                relative_path = os.path.relpath(file_path, folder_path)
+                repo_path = os.path.join(HARDCODED_PATH_IN_REPO, relative_path).replace('\\', '/')
+                
+                print(f"Uploading {file_path} to {repo_path}")
+                
+                api.upload_file(
+                    path_or_fileobj=file_path,
+                    path_in_repo=repo_path,
+                    repo_id=HARDCODED_DATASET_REPO_ID,
+                    repo_type="dataset",
+                    token=HF_TOKEN
+                )
+        
         print("Upload complete!")
     except Exception as e:
         print(f"Upload failed: {str(e)}")
@@ -129,7 +147,7 @@ async def run_transfer():
         if not os.listdir(temp_dir):
             raise Exception("No files were downloaded")
         
-        print(f"Downloaded files: {os.listdir(temp_dir)}")
+        print(f"Final downloaded files: {os.listdir(temp_dir)}")
         
         # Upload to dataset
         await loop.run_in_executor(
@@ -145,7 +163,6 @@ async def run_transfer():
         error_msg = f"Transfer failed: {str(e)}"
         print(error_msg)
         transfer_error = error_msg
-        raise
     finally:
         # Clean up downloaded files
         if temp_dir:
@@ -185,4 +202,4 @@ async def startup_event():
 if __name__ == "__main__":
     import uvicorn
     # Run the server indefinitely
-    uvicorn.run(app, host="0.0.0.0", port=7860)
+    uvicorn.run(app, host="0.0.0.0", port=7860, timeout_keep_alive=600)