Update agents.py

agents.py

@@ -1,18 +1,18 @@
-import os
+#import os
 
 #from langchain.agents import load_tools
-from openai import OpenAI
+#from openai import OpenAI
 #from openinference.instrumentation.smolagents import SmolagentsInstrumentor
 #from phoenix.otel import register
-from smolagents import (
-    CodeAgent,
-    ToolCallingAgent,
-    OpenAIServerModel,
-    Tool,
-    DuckDuckGoSearchTool,
-    WikipediaSearchTool
-)
-from tools import VisitWebpageTool
+#from smolagents import (
+#    CodeAgent,
+#    ToolCallingAgent,
+#    OpenAIServerModel,
+#    Tool,
+#    DuckDuckGoSearchTool,
+#    WikipediaSearchTool
+#)
+#from tools import VisitWebpageTool
 
 ###
 from crewai import Agent
@@ -20,9 +20,9 @@ from langchain_openai import ChatOpenAI
 from tools import scrape_tool, search_tool, today_tool
 ###
 
-MODEL_ID_1 = "gpt-4o-mini"
-MODEL_ID_2 = "gpt-4o"
-MODEL_ID_3 = "o4-mini"
+#MODEL_ID_1 = "gpt-4o-mini"
+#MODEL_ID_2 = "gpt-4o"
+#MODEL_ID_3 = "o4-mini"
 
 #PHOENIX_PROJECT_NAME = "gaia"
 
@@ -38,7 +38,7 @@ MODEL_ID_3 = "o4-mini"
 
 #SmolagentsInstrumentor().instrument(tracer_provider = tracer_provider)
 
-def run_gaia(question, file_name):
+#def run_gaia(question, file_name):
     #search_tool = Tool.from_langchain(load_tools(["serpapi"])[0])
     #wikipedia_tool = Tool.from_langchain(load_tools(["wikipedia"])[0])
     
@@ -62,48 +62,48 @@ def run_gaia(question, file_name):
     #    verbosity_level = 1
     #)
 
-    manager_agent = CodeAgent(
-        #add_base_tools = True,
-        #additional_authorized_imports = ["json", "numpy", "pandas", "time"],
-        #final_answer_checks = [get_final_answer],
-        #managed_agents = [web_search_agent],
-        #max_steps = 5,
-        model = OpenAIServerModel(model_id = MODEL_ID_2),
-        planning_interval=3,
-        tools = [DuckDuckGoSearchTool(), VisitWebpageTool(), WikipediaSearchTool()],
-        verbosity_level = 1
-    )
+    #manager_agent = CodeAgent(
+    #    #add_base_tools = True,
+    #    #additional_authorized_imports = ["json", "numpy", "pandas", "time"],
+    #    #final_answer_checks = [get_final_answer],
+    #    #managed_agents = [web_search_agent],
+    #    #max_steps = 5,
+    #    model = OpenAIServerModel(model_id = MODEL_ID_2),
+    #    planning_interval=3,
+    #    tools = [DuckDuckGoSearchTool(), VisitWebpageTool(), WikipediaSearchTool()],
+    #    verbosity_level = 1
+    #)
 
     #return manager_agent.run(question)
 
-    answer = manager_agent.run(question)
+    #answer = manager_agent.run(question)
 
-    return get_final_answer(question, answer)
+    #return get_final_answer(question, answer)
     
-def get_final_answer(question, answer):
-    prompt_template = """
-        You are an expert in precise question answering. You are given a question and context. You must **precisely** answer the question based on the context and then stop.
-        **Question:** """ + str(question) + """
-        **Context:** """ + str(answer) + """
-        **Example 1:** What is the capital of France? Paris
-        **Example 2:** What is the superlative of good? Best
-        **Example 3:** What is the opposite of left? Right
-        **Answer:**: 
-        """
-
-    client = OpenAI()
-    completion = client.chat.completions.create(
-        messages = [{"role": "user", "content": [{"type": "text", "text": prompt_template}]}],
-        model = MODEL_ID_1
-    )
-
-    final_answer = completion.choices[0].message.content
+#def get_final_answer(question, answer):
+#    prompt_template = """
+#        You are an expert in precise question answering. You are given a question and context. You must **precisely** answer the question based on the context and then stop.
+#        **Question:** """ + str(question) + """
+#        **Context:** """ + str(answer) + """
+#        **Example 1:** What is the capital of France? Paris
+#        **Example 2:** What is the superlative of good? Best
+#        **Example 3:** What is the opposite of left? Right
+#        **Answer:**: 
+#        """
+
+#    client = OpenAI()
+#    completion = client.chat.completions.create(
+#        messages = [{"role": "user", "content": [{"type": "text", "text": prompt_template}]}],
+#        model = MODEL_ID_1
+#    )
+
+#    final_answer = completion.choices[0].message.content
     
-    print(f"Question: {question}")
-    print(f"Answer: {answer}")
-    print(f"Final answer: {final_answer}")
+#    print(f"Question: {question}")
+#    print(f"Answer: {answer}")
+#    print(f"Final answer: {final_answer}")
     
-    return final_answer
+#    return final_answer
 
 ###
 def get_researcher_agent(model, verbose):