Skipped video analysis, gave same tools to llm only agent that multi

agents had, changed model and model provider as together was too
expensive

app.py

@@ -3,6 +3,7 @@ import gradio as gr
 import requests
 import pandas as pd
 import json
+import re
 from phoenix.otel import register
 from openinference.instrumentation.smolagents import SmolagentsInstrumentor
 from llm_only_agent import LLMOnlyAgent
@@ -90,10 +91,11 @@ def run_and_submit_all(profile: gr.OAuthProfile | None, limit: int | None):
     results_log = []
     answers_payload = []
     # Limit for test purposes
-    limit = 2
+    limit = None
     if limit is not None:
         questions_data = questions_data[:limit]
-    print(f"Running agent on {len(questions_data)} questions...")
+    # Run agents on questions fetched
+    print(f"Running agents on {len(questions_data)} questions...")
     for item in questions_data:
         task_id = item.get("task_id")
         file_name = item.get("file_name")
@@ -109,6 +111,9 @@ def run_and_submit_all(profile: gr.OAuthProfile | None, limit: int | None):
         if not task_id or question_text is None:
             print(f"Skipping item with missing task_id or question: {item}")
             continue
+        if re.search("youtube", question_text, re.IGNORECASE):
+            print(f"Skipping item with youtube link: {item}")
+            continue
         try:
             submitted_answer = agent(question_text)
             answers_payload.append(

llm_only_agent.py

@@ -1,12 +1,18 @@
 import re
-from smolagents import AgentMemory, CodeAgent, InferenceClientModel, FinalAnswerTool, WebSearchTool
+from smolagents import (
+    AgentMemory,
+    CodeAgent,
+    InferenceClientModel,
+    FinalAnswerTool,
+    WebSearchTool,
+)
 from collections.abc import Callable
 
 from smolagents.default_tools import VisitWebpageTool, WikipediaSearchTool
 
+
 class LLMOnlyAgent:
     def __init__(self):
-
         # Instructions prompt
         self.instructions = """finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER].
         YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings.
@@ -14,23 +20,34 @@ class LLMOnlyAgent:
         If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise.
         If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string."""
 
-
         # Basic inference model
         model = InferenceClientModel(
-        max_tokens=8096,
-        model_id='Qwen/Qwen2.5-Coder-32B-Instruct',# it is possible that this model may be overloaded
-        custom_role_conversions=None,
+            max_tokens=8096,
+            model_id="Qwen/Qwen3-Coder-30B-A3B-Instruct",
+            custom_role_conversions=None,
+            provider="nebius",
         )
 
         # Code Agent
         self.agent = CodeAgent(
             model=model,
             instructions=self.instructions,
-            tools=[FinalAnswerTool(), WikipediaSearchTool(), WebSearchTool(), VisitWebpageTool()],
-            additional_authorized_imports=[ "markdownify" , "requests" ],
+            tools=[
+                FinalAnswerTool(),
+                WikipediaSearchTool(),
+                WebSearchTool(),
+                VisitWebpageTool(),
+            ],
+            additional_authorized_imports=[
+                "markdownify",
+                "requests",
+                "pandas",
+                "numpy",
+                "chess",
+            ],
             max_steps=5,
-            planning_interval=3
-            #final_answer_checks=self.final_answer_checks()
+            planning_interval=3,
+            # final_answer_checks=self.final_answer_checks()
         )
 
         print("LLM-only Agent initialized.")
@@ -41,10 +58,15 @@ class LLMOnlyAgent:
         print(f"Agent returning answer: {answer}")
         return answer
 
-    def final_answer_checks(self) -> list[Callable] :
-        return [ self.check_func ]
+    def final_answer_checks(self) -> list[Callable]:
+        return [self.check_func]
 
     def check_func(self, answer: str, memory: AgentMemory) -> bool:
-        check = bool(re.match(r'^(\d+(\.\d+)?|\w+(\s+\w+){0,4}|(\d+(\.\d+)?|"[^"]*"|\w+)(\s*,\s*(\d+(\.\d+)?|"[^"]*"|\w+))+)$', answer))
+        check = bool(
+            re.match(
+                r'^(\d+(\.\d+)?|\w+(\s+\w+){0,4}|(\d+(\.\d+)?|"[^"]*"|\w+)(\s*,\s*(\d+(\.\d+)?|"[^"]*"|\w+))+)$',
+                answer,
+            )
+        )
         print(f"FINAL ANSWER CHECK is {check}")
         return check