Update app.py

app.py

@@ -1,177 +1,23 @@
-import os, sys
-from enum import Enum
+import os
 import gradio as gr
 import requests
 import inspect
-import subprocess
-import dateparser
-from bs4 import BeautifulSoup
-import regex
 import pandas as pd
-import torch
-from functools import lru_cache
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
-from smolagents import CodeAgent, WebSearchTool, WikipediaSearchTool, VisitWebpageTool, PythonInterpreterTool
-import smolagents.tools as _tools
-from smolagents.models import ChatMessage
-# from huggingface_hub import InferenceClient, hf_hub_download
-
-subprocess.run(["playwright", "install"], check=True)
-
-print(dir(_tools))
 
 # (Keep Constants as is)
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# class LocalLLM:
-#     def __init__(self, pipe):
-#         self.pipe = pipe
-
-#     def generate(self, prompt, **kwargs):
-#         unsupported_keys = ["stop_sequences"] # Remove keys not accepted by HF pipelines
-#         cleaned_kwargs = {k: v for k, v in kwargs.items() if k not in unsupported_keys}
-#         # print(f"🧪 kwargs cleaned: {cleaned_kwargs.keys()}")
-#         try:
-#             outputs = self.pipe(prompt, **cleaned_kwargs)
-#             # print(f"🧪 Raw output from pipe: {outputs}")
-#             if isinstance(outputs, list) and isinstance(outputs[0], dict):
-#                 out = outputs[0]["generated_text"]
-#             elif isinstance(outputs, list):
-#                 out = outputs[0]  # fallback if it's just a list of strings
-#             else:
-#                 out = str(outputs)
-#             print("🧪 Final object to return:", type(out), out[:100])
-#             return {'role': 'assistant', 'content': [{'type':'text', 'text': out}]}
-#         except Exception as e:
-#             print(f"❌ Error in LocalLLM.generate(): {e}")
-#             raise
-
-def check_token_access():
-    token = os.environ.get("HF_TOKEN", "")
-    if not token:
-        print("❌ No token found")
-        return
-    headers = {"Authorization": f"Bearer {token}"}
-    url = "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct/resolve/main/config.json"
-    try:
-        r = requests.get(url, headers=headers, timeout=10)
-        print(f"🔍 Token test response: {r.status_code}")
-        if r.status_code == 200:
-            print("✅ Token access confirmed for gated model.")
-        elif r.status_code == 403:
-            print("❌ 403 Forbidden: Token does not have access.")
-        else:
-            print("⚠️ Unexpected status:", r.status_code)
-    except Exception as e:
-        print("❌ Token check failed:", e)
-
-class CachedWebSearchTool(WebSearchTool):
-    @lru_cache(maxsize=128)
-    def run(self, query: str):
-        # identical queries return instantly
-        return super().run(query)
-
-class CachedWikiTool(WikipediaSearchTool):
-    @lru_cache(maxsize=128)
-    def run(self, page: str):
-        return super().run(page)
-
 # --- Basic Agent Definition ---
-# ----- THIS IS WHERE YOU CAN BUILD WHAT YOU WANT ------
+# ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 class BasicAgent:
-    def __init__(self, model_id="meta-llama/Llama-3.1-8B-Instruct", hf_token=""):
+    def __init__(self):
         print("BasicAgent initialized.")
-        print("ENV-HF_TOKEN-LEN", len(hf_token), file=sys.stderr)
-        check_token_access()
-
-        # Local test
-        # client = InferenceClient(
-        # model="meta-llama/Llama-3.1-8B-Instruct",
-        # token=os.environ["HF_TOKEN"]
-        # )
-        # print(client.text_generation("Hello, my name is", max_new_tokens=20))
-        
-        # Initialize the model
-        # model = HfApiModel(model_id="meta-llama/Llama-3.1-8B-Instruct",
-        #                   # format="text-generation",
-        #                   token=os.environ["HF_TOKEN"], 
-        #                   max_tokens=2048,
-        #                   temperature=0.0
-        #                  )
-        
-        # Initialize the tools other than the base tools
-        # See list of base tools in https://github.com/huggingface/smolagents/blob/main/src/smolagents/default_tools.py
-        
-        # Download the model weights and build the pipeline
-        tok = AutoTokenizer.from_pretrained(model_id, token=hf_token)
-        mod = AutoModelForCausalLM.from_pretrained(
-            model_id,
-            torch_dtype=torch.float16,
-            device_map="auto",  # auto-distributes to GPU
-            token=hf_token
-        )
-        self.pipe = pipeline(
-            "text-generation",
-            model=mod,
-            tokenizer=tok,
-            max_new_tokens=512,
-            return_full_text=False,      # <— only get the completion, not the prompt + completion
-            # temperature=1.0,
-        )
-        # Introduce tools
-        wiki_tool = CachedWikiTool()
-        search_tool = CachedWebSearchTool()
-        python_tool = PythonInterpreterTool()
-        html_parse_tool = VisitWebpageTool()
-        # Initialize the agent
-        self.agent = CodeAgent(model=self, 
-                               tools=[wiki_tool, search_tool, python_tool, html_parse_tool], 
-                               add_base_tools=True,
-                               additional_authorized_imports=["dateparser", "bs4", "regex"])
-        
-    def _serialize_messages(self, messages):
-        prompt = []
-        for m in messages:
-            r = m["role"]
-            role = r.value if isinstance(r, Enum) and hasattr(r, "value") else r # "system" / "user" / "assistant"
-            text = "".join([c['text'] for c in m['content']])
-            prompt.append(f"{role}: {text}")
-        return "\n".join(prompt)
-
-    def generate(self, question: str, stop_sequences=None, **kwargs) -> str:
+    def __call__(self, question: str) -> str:
         print(f"Agent received question (first 50 chars): {question[:50]}...")
-        # 1. Build the HF kwargs
-        allowed = {"max_new_tokens", "temperature", "top_k", "top_p"}
-        gen_kwargs = {k: v for k, v in kwargs.items() if k in allowed}
-        
-        # 2. Serialize the message and get the response
-        prompt_str = (
-            self._serialize_messages(question)
-            if isinstance(question, list)
-            else question
-        )
-        outputs = self.pipe(prompt_str, **gen_kwargs)
-        response = outputs[0]["generated_text"]
-        # response = self.agent.run(question)
-
-        # 3. Optionally map SmolAgents’ stop_sequences → HF pipeline’s 'stop'
-        if stop_sequences:
-            # find the earliest occurrence of any stop token
-            cuts = [response.find(s) for s in stop_sequences if response.find(s) != -1]
-            if cuts:
-                response = response[: min(cuts)]
-        
-        print(f"Agent returning its generated answer: {response}")
-        
-        # wrap back into a chat message dict
-        return ChatMessage(role="assistant", content=response)
-        # return {
-        #     "role": 'assistant',
-        #     "content": [{"type": "text", "text": response}],
-        # }
-
-    __call__ = generate
+        fixed_answer = "This is a default answer."
+        print(f"Agent returning fixed answer: {fixed_answer}")
+        return fixed_answer
 
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
@@ -180,7 +26,6 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
     # --- Determine HF Space Runtime URL and Repo URL ---
     space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
-    hf_token = os.getenv("HF_TOKEN")
 
     if profile:
         username= f"{profile.username}"
@@ -195,7 +40,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
 
     # 1. Instantiate Agent ( modify this part to create your agent)
     try:
-        agent = BasicAgent(hf_token=hf_token).agent
+        agent = BasicAgent()
     except Exception as e:
         print(f"Error instantiating agent: {e}")
         return f"Error initializing agent: {e}", None
@@ -224,8 +69,6 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
         print(f"An unexpected error occurred fetching questions: {e}")
         return f"An unexpected error occurred fetching questions: {e}", None
 
-    questions_data = questions_data[:5]
-    
     # 3. Run your Agent
     results_log = []
     answers_payload = []