Merge pull request #112 from joshuasundance-swca/pyhistory

langchain-streamlit-demo/app.py

@@ -430,7 +430,7 @@ if st.session_state.llm:
                     callbacks=callbacks,
                     tags=["Streamlit Chat"],
                     verbose=True,
-                    return_intermediate_steps=True,
+                    return_intermediate_steps=False,
                 )
                 if st.session_state.provider == "Anthropic":
                     config["max_concurrency"] = 5

langchain-streamlit-demo/llm_resources.py

@@ -3,9 +3,6 @@ from tempfile import NamedTemporaryFile
 from typing import Tuple, List, Optional, Dict
 
 from langchain.agents import AgentExecutor
-from langchain.agents.openai_functions_agent.agent_token_buffer_memory import (
-    AgentTokenBufferMemory,
-)
 from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
 from langchain.callbacks.base import BaseCallbackHandler
 from langchain.chains import LLMChain
@@ -57,10 +54,17 @@ def get_agent(
     )
     agent = OpenAIFunctionsAgent(llm=llm, tools=tools, prompt=prompt)
 
-    agent_memory = AgentTokenBufferMemory(
+    # agent_memory = AgentTokenBufferMemory(
+    #     chat_memory=chat_history,
+    #     memory_key=memory_key,
+    #     llm=llm,
+    # )
+    from langchain.memory import ConversationBufferMemory
+
+    agent_memory = ConversationBufferMemory(
         chat_memory=chat_history,
+        return_messages=True,
         memory_key=memory_key,
-        llm=llm,
     )
 
     agent_executor = AgentExecutor(
@@ -68,7 +72,7 @@ def get_agent(
         tools=tools,
         memory=agent_memory,
         verbose=True,
-        return_intermediate_steps=True,
+        return_intermediate_steps=False,
         callbacks=callbacks,
     )
     return (

langchain-streamlit-demo/python_coder.py

@@ -1,16 +1,14 @@
 """langchain python coder-- requires black, ruff, and mypy."""
 
-import argparse
 import os
 import re
 import subprocess  # nosec
 import tempfile
-from importlib.util import find_spec
 
 from langchain.agents import initialize_agent, AgentType
 from langchain.agents.tools import Tool
-from langchain.chat_models import ChatOpenAI
 from langchain.llms.base import BaseLLM
+from langchain.memory import ConversationBufferMemory
 from langchain.prompts import ChatPromptTemplate
 from langchain.pydantic_v1 import BaseModel, validator, Field, ValidationError
 
@@ -177,35 +175,17 @@ def get_agent(
     llm: BaseLLM,
     agent_type: AgentType = AgentType.CHAT_CONVERSATIONAL_REACT_DESCRIPTION,
 ):
+    agent_memory = ConversationBufferMemory(
+        return_messages=True,
+        memory_key="chat_history",
+    )
     return initialize_agent(
         tools,
         llm,
         agent=agent_type,
         verbose=True,
         handle_parsing_errors=True,
+        memory=agent_memory,
         prompt=prompt,
-        # return_intermediate_steps=True,
+        return_intermediate_steps=False,
     ) | (lambda output: output["output"])
-
-
-if __name__ == "__main__":
-    for lib in ("black", "ruff", "mypy"):
-        if find_spec(lib) is None:
-            raise ImportError(
-                "You must install black, ruff, and mypy to use this tool. "
-                "You can install them with `pip install black ruff mypy`.",
-            )
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--model", "-m", default="gpt-4-1106-preview")
-    parser.add_argument("instruction")
-    args = parser.parse_args()
-
-    instruction = args.instruction
-    model = args.model
-
-    llm = ChatOpenAI(model_name=model, temperature=0.0)
-    agent = get_agent(llm)
-
-    output = agent.invoke({"input": instruction}, config=dict(verbose=True))
-    print(output)