langchain-ai / streamlit-agent Goto Github PK

Hello, I tried to run 🦜 LangChain: Chat with Documents from repo and was getting this error and some other error like List has no attribute Embedding. I am just wondering if is there a bug within Langchain that requires update/upgrade? Both title Error as well the descriptive Error here in the body message have been experienced by me today.

Hey there 👋 Just wanted to let you know that your app on Streamlit Cloud deployed from this repo has gone over its resource limits. Access to the app is temporarily limited. Visit the app to see more details and possible solutions.

[tool.poetry]
name = "streamlit-agent"
version = "0.1.0"
description = ""
authors = ["Tim Conkling [email protected]", "Joshua Carroll [email protected]"]
license = "Apache 2.0"
readme = "README.md"
packages = [{include = "streamlit_agent"}]

[tool.poetry.dependencies]
python = ">=3.10,<4.0"
langchain = {version = ">=0.0.252", extras = ["docarray"]}
openai = "^0.27.8"
duckduckgo-search = "^3.8.3"
pypdf = "^3.12.2"
sentence-transformers = "^2.2.2"
torch = ">=2.0.0, !=2.0.1"
tabulate = "^0.9.0"
streamlit-feedback = "^0.0.9"
langchain-experimental = "^0.0.10"
streamlit= ">=1.26"

[tool.poetry.group.dev.dependencies]
black = "^23.3.0"
mypy = "^1.4.1"
pre-commit = "^3.3.3"

[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"

PS C:\Users\39627\Downloads\streamlit-agent-main> & c:/Users/39627/Downloads/streamlit-agent-main/langchain_env/Scripts/python.exe c:/Users/39627/Downloads/streamlit-agent-main/streamlit_agent/chat_with_documents.py
Traceback (most recent call last):
File "c:\Users\39627\Downloads\streamlit-agent-main\streamlit_agent\chat_with_documents.py", line 4, in
from langchain.chat_models import ChatOpenAI
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain_init_.py", line 6, in
from langchain.agents import MRKLChain, ReActChain, SelfAskWithSearchChain
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\agents_init_.py", line 31, in
from langchain.agents.agent import (
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\agents\agent.py", line 14, in
from langchain.agents.agent_iterator import AgentExecutorIterator
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\agents\agent_iterator.py", line 21, in
from langchain.callbacks.manager import (
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\callbacks_init_.py", line 10, in
from langchain.callbacks.aim_callback import AimCallbackHandler
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\callbacks\aim_callback.py", line 5, in
from langchain.schema import AgentAction, AgentFinish, LLMResult
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\schema_init_.py", line 3, in
from langchain.schema.cache import BaseCache
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\schema\cache.py", line 6, in
from langchain.schema.output import Generation
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\schema\output.py", line 7, in
from langchain.load.serializable import Serializable
File "C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\load\serializable.py", line 4, in
from langchain.pydantic_v1 import BaseModel, PrivateAttr
ImportError: cannot import name 'PrivateAttr' from 'langchain.pydantic_v1' (C:\Users\39627\Downloads\streamlit-agent-main\langchain_env\Lib\site-packages\langchain\pydantic_v1_init_.py)

Hitting this error:

Traceback (most recent call last):
  File "/Users/josoroma/Library/Caches/pypoetry/virtualenvs/etl-Fd0mW_QZ-py3.11/lib/python3.11/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 552, in _run_script
    exec(code, module.__dict__)
  File "/Users/josoroma/sites/etl-pipeline-for-langchain-docs/chat.py", line 111, in <module>
    ChatApp().main()
  File "/Users/josoroma/sites/etl-pipeline-for-langchain-docs/chat.py", line 83, in main
    for msg in st.session_state['messages']:
               ~~~~~~~~~~~~~~~~^^^^^^^^^^^^
  File "/Users/josoroma/Library/Caches/pypoetry/virtualenvs/etl-Fd0mW_QZ-py3.11/lib/python3.11/site-packages/streamlit/runtime/state/session_state_proxy.py", line 90, in __getitem__
    return get_session_state()[key]
           ~~~~~~~~~~~~~~~~~~~^^^^^
  File "/Users/josoroma/Library/Caches/pypoetry/virtualenvs/etl-Fd0mW_QZ-py3.11/lib/python3.11/site-packages/streamlit/runtime/state/safe_session_state.py", line 111, in __getitem__
    raise KeyError(key)
KeyError: 'messages'

My code:

import os
import weaviate
import openai
from pydantic import BaseModel
from langchain.callbacks.base import BaseCallbackHandler
# from langchain.chat_models import ChatOpenAI
from langchain.schema import ChatMessage
import streamlit as st

class Document(BaseModel):
    content: str

class QueryResult(BaseModel):
    document: Document

class StreamHandler(BaseCallbackHandler):
    def __init__(self, container, initial_text=""):
        self.container = container
        self.text = initial_text

    def on_llm_new_token(self, token: str, **kwargs) -> None:
        self.text += token
        self.container.markdown(self.text)

class ChatApp:
    def __init__(self):
        self.client = None
        self.get_env_variables()
        self.client = self.get_client()

    def get_env_variable(self, var_name):
        var_value = os.getenv(var_name)
        return var_value

    def get_env_variables(self):
        with st.sidebar:
            self.OPENAI_API_KEY = self.get_env_variable("OPENAI_API_KEY") or st.text_input("OpenAI API Key", type="password")
            self.WEAVIATE_HOST = self.get_env_variable("WEAVIATE_HOST") or st.text_input("Weaviate Host")
            self.WEAVIATE_AUTH_API_KEY = self.get_env_variable("WEAVIATE_AUTH_API_KEY") or st.text_input("Bearer Token", type="password")

        if not self.OPENAI_API_KEY or not self.WEAVIATE_HOST or not self.WEAVIATE_AUTH_API_KEY:
            st.info("Please add your OpenAI API Key, Weaviate Host, and Bearer Token to continue.")
            st.stop()

        openai.api_key = self.OPENAI_API_KEY

    def get_client(self):
        try:
            client = weaviate.Client(
                url=self.WEAVIATE_HOST,
                auth_client_secret=weaviate.AuthApiKey(self.WEAVIATE_AUTH_API_KEY),
                additional_headers={"X-OpenAI-Api-Key": self.OPENAI_API_KEY},
            )
        except Exception as e:
            st.error(f"Error occurred while creating the Weaviate client: {str(e)}")
            st.stop()

        return client

    def client_query(self, question: str):
        generatePrompt = "Respond to the human as helpfully and accurately as possible: {text}"
        nearText = {"concepts": [f"{question}"]}

        try:
            response = (
                self.client.query
                .get("Document", ["content"])
                .with_generate(single_prompt=generatePrompt)
                .with_near_text(nearText)
                .with_limit(1)
                .do()
            )
        except Exception as e:
            st.error(f"Error occurred while querying the Weaviate client: {str(e)}")
            st.stop()

        return response

    def main(self):
        if 'messages' not in st.session_state:
            st.session_state['messages'] = [ChatMessage(role="assistant", content="How can I help you?")]

        for msg in st.session_state['messages']:
            st.chat_message(msg.role).write(msg.content)

        if prompt := st.text_input("Your input:"):
            st.session_state['messages'].append(ChatMessage(role="user", content=prompt))
            st.chat_message("user").write(prompt)

            response = self.client_query(prompt)
            if response:
                try:
                    for document in response['data']['Get']['Document']:
                        try:
                            generativeOpenAI = document['_additional']['generate']["singleResult"]
                            content = document['content']
                        except KeyError as ke:
                            st.markdown(f"Error: Expected keys not found in the document. {ke}")
                            continue

                        if generativeOpenAI:
                            st.session_state['messages'].append(ChatMessage(role="assistant", content=generativeOpenAI))
                            st.chat_message("assistant").write(generativeOpenAI)
                        if content:
                            st.session_state['messages'].append(ChatMessage(role="assistant", content=content))
                            st.chat_message("assistant").write(content)
                except KeyError as ke:
                    st.markdown(f"Error: Expected keys not found in the response. {ke}")

if __name__ == "__main__":
    ChatApp().main()

I have several times, but its not working!

When running chat_with_documents.py , I get the following error message

Error in PrintRetrievalHandler.on_retriever_start callback: PrintRetrievalHandler.on_retriever_start() takes 2 positional arguments but 3 were given

Apparently, it still works fine, and PrintRetrievalHandler.on_retriever_start() can be commented out

See the error below.

openai account has rate limit. The application will output error.
anyway to fix this issue?

I love this repo, and is very useful. Although im wondering if there's an implementation to prompt the user for the "human" tool. Cause as of now, everything happens in the background, and when the human is "prompted" nothing happens.

I have the problem that docarray is not compatible with pydantic as i get always the error
ImportERROR: cannot import "ROOT_KEY" from "pydantic.main"

i tried to downgrade the pydantic to pydantic 1.10.12 but it still not working any one ahve an idea how to solve this problem?

Thanks for your amazing work.

`class PrintRetrievalHandler(BaseCallbackHandler):
def init(self, container):
self.status = container.status("Context Retrieval")

def on_retriever_start(self, serialized: dict, query: str, **kwargs):
    self.status.write(f"**Question:** {query}")
    self.status.update(label=f"**Context Retrieval:** {query}")

def on_retriever_end(self, documents, **kwargs):
    for idx, doc in enumerate(documents):
        source = os.path.basename(doc.metadata["source"])
        self.status.write(f"**Document {idx} from {source}**")
        self.status.markdown(doc.page_content)
    self.status.update(state="complete")`

Here a status container is initialized. But the last one will disappear when starting a new retrieval. I know in streamlit, the session_state is used to store the objects, but I am new in streamlit, and I failed to implement.

Thanks for your attention. I appreciate it a lot.

https://langchain-chat-search.streamlit.app/ is still broken:

Dear Developer,

Thanks for sharing your work. However the below streamlit agent doesn't work, can pls have a check on it? thank you.
https://langchain-document-chat.streamlit.app/

regards,
Sun Yan

I was trying to use the app just for demo purpose and to see how it goes. Used my own openai api key and in the text box said 'describe the first table'. This is what I get. Not sure what is the issue.

Do we really need it, I don't see explicit use of these rather adding extra space and time to create an environment.

sentence-transformers = "^2.2.2"
torch = ">=2.0.0, !=2.0.1"

I'm running

https://github.com/langchain-ai/streamlit-agent/blob/main/streamlit_agent/chat_with_documents.py

and it works seamlessly with my private account and ChatOpenAI chat_llm

if you keep everything the same, but switch to AzureChatOpenAI chat_llm,

then the history start to misbehave, bot reply vanishes, etc.

just tried the above mentioned two examples with my own freshly created openai api key and got error messages. one was:

duckduckgo_search.exceptions.DuckDuckGoSearchException: This app has encountered an error. The original error message is redacted to prevent data leaks. Full error details have been recorded in the logs (if you're on Streamlit Cloud, click on 'Manage app' in the lower right of your app). [...]

I.e. the intermediate thoughts appear in the most recent answer from the agent. However, when more messages are added, the intermediate thoughts disappear from the older answers (leaving just the final answer). I would like to be able leave the intermediate thoughts in place, to allow going back and inspecting them.

When installing dependencies from the lock file with poetry, M1 users will get clang: error: the clang compiler does not support '-march=native'

I found a workaround by setting an environment variable (Ref:https://stackoverflow.com/a/75870031/12906920)
export HNSWLIB_NO_NATIVE=1
then
poetry install

Just wanted to inform this issue.

Thanks.