Fix gradio version conflict

agent.py

@@ -6,6 +6,7 @@ from typing import Optional
 from prompts import build_solver_prompt
 from tools import TaskFileTool
 from utils import extract_final_answer, normalize_final_answer
+from llm_client import HFLLMClient
 
 
 @dataclass
@@ -24,12 +25,11 @@ class SubmissionAgent:
     - Return ONLY the final answer string
     - Stay framework-agnostic for now so we can plug in any LLM later
     """
-
-    def __init__(self, llm_client=None, config: Optional[AgentConfig] = None):
-        self.llm_client = llm_client
+    def __init__(self, config: Optional[AgentConfig] = None, llm_client=None):
         self.config = config or AgentConfig()
+        self.llm_client = llm_client or HFLLMClient()
         self.task_file_tool = TaskFileTool(api_base_url=self.config.api_base_url)
-
+        
     def __call__(self, question: str, task_id: Optional[str] = None) -> str:
         """
         Main entry point used by app.py.
@@ -67,12 +67,9 @@ class SubmissionAgent:
         """
         prompt = build_solver_prompt(question=question, context=context)
 
-        if self.llm_client is None:
-            # Safe placeholder so the app can run while we build the stack.
-            # We will replace this with a real model client later.
-            return "PLACEHOLDER"
-
         try:
             return self.llm_client.generate(prompt)
-        except Exception:
-            return "PLACEHOLDER"
+        except Exception as e:
+            print(f"LLM generation error: {e}")
+            return ""
+            

app.py

@@ -187,7 +187,6 @@ with gr.Blocks() as demo:
 
     run_button.click(
         fn=run_and_submit_all,
-        inputs=[login_button],
         outputs=[status_output, results_table],
     )
 

llm_client.py

@@ -0,0 +1,38 @@
+import os
+from huggingface_hub import InferenceClient
+from dotenv import load_dotenv
+
+load_dotenv()
+
+
+class HFLLMClient:
+    def __init__(self):
+        self.api_key = os.getenv("HF_TOKEN")
+        print("HF token present:", bool(self.api_key))
+
+        if not self.api_key:
+            raise ValueError("HF_TOKEN is not set")
+
+        self.model = "Qwen/Qwen2.5-7B-Instruct"
+        self.client = InferenceClient(
+            provider="auto",
+            api_key=self.api_key,
+        )
+
+    def generate(self, prompt: str) -> str:
+        try:
+            output = self.client.chat_completion(
+                model=self.model,
+                messages=[
+                    {"role": "user", "content": prompt}
+                ],
+                max_tokens=64,
+                temperature=0.1,
+            )
+
+            text = output.choices[0].message.content
+            print("LLM response preview:", str(text)[:300])
+            return str(text)
+
+        except Exception as e:
+            raise ValueError(f"Inference call failed: {e}")