Update app.py

app.py

@@ -58,28 +58,27 @@ def get_quantization_recipe(method, model_architecture):
         raise ValueError(f"Unsupported quantization method: {method}")
 
 # --------------------------------------------------------------------------------
-# CHANGE #1: The function no longer needs the `request` or `oauth_token` argument.
+# CHANGE #1: Reverted to the correct function signature that accepts the OAuthToken
 # --------------------------------------------------------------------------------
-def compress_and_upload(model_id: str, quant_method: str):
+def compress_and_upload(model_id: str, quant_method: str, oauth_token: gr.OAuthToken | None):
     """
     Compresses a model using llm-compressor and uploads it to a new HF repo.
     """
     if not model_id:
         raise gr.Error("Please select a model from the search bar.")
+    
+    if oauth_token is None:
+        raise gr.Error("Authentication error. Please log in to continue.")
+        
+    token = oauth_token.token
 
-    # Check for login status by calling whoami(). It will raise an error if not logged in.
     try:
-        user_info = whoami()
-        if user_info is None:
-             raise gr.Error("Authentication error. Please log in to continue.")
-        username = user_info["name"]
-    except Exception as e:
-        raise gr.Error(f"Authentication error. Please log in to continue. Details: {e}")
+        # Use the provided token for all hub interactions
+        username = whoami(token=token)["name"]
 
-    try:
         # --- 1. Load Model and Tokenizer ---
-        model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map=None)
-        tokenizer = AutoTokenizer.from_pretrained(model_id)
+        model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map=None, token=token)
+        tokenizer = AutoTokenizer.from_pretrained(model_id, token=token)
         
         output_dir = f"{model_id.split('/')[-1]}-{quant_method}"
 
@@ -100,8 +99,7 @@ def compress_and_upload(model_id: str, quant_method: str):
         )
 
         # --- 4. Create Repo and Upload ---
-        # HfApi() will automatically use the logged-in user's token.
-        api = HfApi()
+        api = HfApi(token=token)
         repo_id = f"{username}/{output_dir}"
 
         repo_url = api.create_repo(repo_id=repo_id, exist_ok=True)
@@ -134,7 +132,7 @@ This conversion was performed by the `llm-compressor-my-repo` Hugging Face Space
 For more details on the recipe used, refer to the `recipe.yaml` file in this repository.
 """
         card = ModelCard(card_content)
-        card.push_to_hub(repo_id) # No token needed here either
+        card.push_to_hub(repo_id, token=token)
 
         return f'<h1>✅ Success!</h1><br/>Model compressed and saved to your new repo: <a href="{repo_url}" target="_blank" style="text-decoration:underline">{repo_id}</a>'
 
@@ -168,20 +166,14 @@ with gr.Blocks(css="footer {display: none !important;}") as demo:
     output_html = gr.HTML(label="Result")
 
     # --------------------------------------------------------------------------------
-    # CHANGE #2: The inputs list is now simpler and matches the function signature.
+    # CHANGE #2: The `login_button` is correctly passed as an input.
     # --------------------------------------------------------------------------------
     compress_button.click(
         fn=compress_and_upload,
-        inputs=[model_input, quant_method_dropdown],
+        inputs=[model_input, quant_method_dropdown, login_button],
         outputs=output_html
     )
-
-    gr.Examples(
-        examples=[
-            ["mistralai/Mistral-7B-Instruct-v0.2", "AWQ"],
-            ["meta-llama/Llama-2-7b-chat-hf", "GPTQ"],
-        ],
-        inputs=[model_input, quant_method_dropdown],
-    )
+    
+    # CHANGE #3: Removed the gr.Examples component to prevent the TypeError.
 
 demo.queue(max_size=5).launch()