Datasets:

delayedkarma

/

langchain-issues

like 2

https://api.github.com/repos/langchain-ai/langchain/issues/15000

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/15000/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/15000/comments

https://api.github.com/repos/langchain-ai/langchain/issues/15000/events

https://github.com/langchain-ai/langchain/issues/15000

I_kwDOIPDwls56TksW

Issue: not getting output as per prompt,what is neccessary changes i need to do?

open

NONE

### Issue you'd like to raise. I am not getting output as per prompt, what is neccessary modification or changes i need to do. def chat_langchain(new_project_qa, query, not_uuid): check = query.lower() relevant_document = result['source_documents'] user_experience_inst = UserExperience.objects.get(not_uuid=not_uuid) custom_prompt_template = f"""You are a Chatbot answering questions. Use the following pieces of context to answer the question at the end. If you don't know the answer, say that you don't know, don't try to make up an answer. {relevant_document} Question: {check} Helpful Answer:""" CUSTOMPROMPT = PromptTemplate( template=custom_prompt_template, input_variables=["context", "question"] ) print(CUSTOMPROMPT,"------------------") new_project_qa.combine_documents_chain.llm_chain.prompt = CUSTOMPROMPT result = new_project_qa(query) if relevant_document: source = relevant_document[0].metadata.get('source', '') # Check if the file extension is ".pdf" file_extension = os.path.splitext(source)[1] if file_extension.lower() == ".pdf": source = os.path.basename(source) # Retrieve the UserExperience instance using the provided not_uuid user_experience_inst = UserExperience.objects.get(not_uuid=not_uuid) bot_ending = user_experience_inst.bot_ending_msg if user_experience_inst.bot_ending_msg is not None else "" # Create the list_json dictionary if bot_ending != '': list_json = { 'bot_message': result['result'] + '\n\n' + str(bot_ending), "citation": source } else: list_json = { 'bot_message': result['result'] + str(bot_ending), "citation": source } else: # Handle the case when relevant_document is empty list_json = { 'bot_message': result['result'], 'citation': '' } # Return the list_json dictionary return list_json ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/15000/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14999

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14999/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14999/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14999/events

https://github.com/langchain-ai/langchain/pull/14999

PR_kwDOIPDwls5ii1-6

fixed wrong link in documentation

closed

CONTRIBUTOR

<!-- Thank you for contributing to LangChain! Please title your PR "<package>: <description>", where <package> is whichever of langchain, community, core, experimental, etc. is being modified. Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes if applicable, - **Dependencies:** any dependencies required for this change, - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` from the root of the package you've modified to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/docs/integrations` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. --> See #14998

https://api.github.com/repos/langchain-ai/langchain/issues/14999/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14998

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14998/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14998/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14998/events

https://github.com/langchain-ai/langchain/issues/14998

I_kwDOIPDwls56TXim

DOC: Document PAGE NOT FOUND

closed

NONE

### Issue with current documentation: 404 page: https://python.langchain.com/docs/contributing/integration referenced by: https://python.langchain.com/docs/contributing/  ### Idea or request for content: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14998/timeline

completed

https://api.github.com/repos/langchain-ai/langchain/issues/14997

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14997/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14997/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14997/events

https://github.com/langchain-ai/langchain/pull/14997

PR_kwDOIPDwls5iidQ4

community: fix for surrealdb client 0.3.2 update + store and retrieve metadata

closed

CONTRIBUTOR

Surrealdb client changes from 0.3.1 to 0.3.2 broke the surrealdb vectore integration. This PR updates the code to work with the updated client. The change is backwards compatible with previous versions of surrealdb client. Also expanded the vector store implementation to store and retrieve metadata that's included with the document object.

https://api.github.com/repos/langchain-ai/langchain/issues/14997/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14996

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14996/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14996/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14996/events

https://github.com/langchain-ai/langchain/issues/14996

I_kwDOIPDwls56SoCp

Add implementation of '_create_chat_result()' method for MiniMax's current implementation

open

NONE

### Feature request Add the implementation of '_create_chat_result()' method for MiniMax's current implementation to allow it being accepted as one of the chat models ### Motivation Currently MiniMax's chat functionality does not work properly with Langchain, as described in this issue: https://github.com/langchain-ai/langchain/issues/14796 The investigation to this bug suggests a missing implementation of method '_create_chat_result()'. With a proper implementation of this method, the `_generate` method will be able to return `ChatResult` objects instead of unaccepted `str`. ### Your contribution I am currently investigating on how to implement it myself, and I am happy to provide any support, including discussion, testing, etc.

https://api.github.com/repos/langchain-ai/langchain/issues/14996/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14995

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14995/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14995/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14995/events

https://github.com/langchain-ai/langchain/pull/14995

PR_kwDOIPDwls5ih8n6

Update Ollama multi-modal multi-vector template README.md

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14995/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14994

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14994/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14994/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14994/events

https://github.com/langchain-ai/langchain/pull/14994

PR_kwDOIPDwls5ih8hh

Update Ollama multi-modal template README.md

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14994/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14993

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14993/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14993/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14993/events

https://github.com/langchain-ai/langchain/pull/14993

PR_kwDOIPDwls5ih8Yw

Update Gemini template README.md

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14993/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14992

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14992/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14992/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14992/events

https://github.com/langchain-ai/langchain/pull/14992

PR_kwDOIPDwls5ih8T9

Update multi-modal multi-vector template README.md

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14992/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14991

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14991/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14991/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14991/events

https://github.com/langchain-ai/langchain/pull/14991

PR_kwDOIPDwls5ih8Gl

Update multi-modal template README.md

closed

COLLABORATOR

<!-- Thank you for contributing to LangChain! Please title your PR "<package>: <description>", where <package> is whichever of langchain, community, core, experimental, etc. is being modified. Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes if applicable, - **Dependencies:** any dependencies required for this change, - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` from the root of the package you've modified to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/docs/integrations` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. -->

https://api.github.com/repos/langchain-ai/langchain/issues/14991/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14990

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14990/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14990/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14990/events

https://github.com/langchain-ai/langchain/pull/14990

PR_kwDOIPDwls5ih2Wv

community[minor] langchain[patch]: convert class tool to openai function

open

NONE

Covert tools to a JSON format description for Openai. Previously, the function `format_tool_to_openai_function` does not return the correct parameter name when given a tool that subclassing the BaseTool: ```python import os from langchain.agents import load_tools from langchain.tools.render import format_tool_to_openai_function os.environ['SERPER_API_KEY'] = '1' * 40 tools = load_tools(["google-serper"]) format_tool_to_openai_function(tools[0])['parameters'] # {'properties': {'__arg1': {'title': '__arg1', 'type': 'string'}}, 'required': ['__arg1'], 'type': 'object'} ``` This patch is trying to fix it. The correct output should be `{'properties': {'query': {'title': 'Query', 'type': 'string'}}, 'required': ['query'], 'type': 'object'}`

https://api.github.com/repos/langchain-ai/langchain/issues/14990/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14989

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14989/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14989/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14989/events

https://github.com/langchain-ai/langchain/pull/14989

PR_kwDOIPDwls5ih0vK

community[patch]: Fix ip relevance score

open

NONE

fix method _max_inner_product_relevance_score_fn when DistanceStrategy is MAX_INNER_PRODUCT [#14948](https://github.com/langchain-ai/langchain/issues/14948)

https://api.github.com/repos/langchain-ai/langchain/issues/14989/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14988

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14988/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14988/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14988/events

https://github.com/langchain-ai/langchain/issues/14988

I_kwDOIPDwls56SQdC

Issue: Adding function calling to ConversationalRetrievalChain

open

NONE

### Issue you'd like to raise. I am having a problem adding function calling to a chain of type ConversationalRetrievalChain. I need help finding a solution. Here is my code, which creates a ConversationalRetrievalChain to retrieve local knowledge and generate chat history information in a summary format. It works fine. However, when I try to add a call to weather_function, I don't know where to add it. I have browsed most of the documentation and couldn't find a solution. Can anyone help me? Thank you! ```python documents = TextLoader("./file/text.txt").load() text_splitter = CharacterTextSplitter(chunk_size=300, chunk_overlap=50) docs = text_splitter.split_documents(documents) embeddings = OpenAIEmbeddings(openai_api_key=APIKEY, openai_api_base=OPENAI_API_BASE) db = FAISS.from_documents(docs, embeddings) retriever = db.as_retriever() Template = """You are a good man and happy to chat with everyone: {context} history chat information in summary: {chat_history} Question: {question} """ prompt = PromptTemplate( input_variables=["context", "chat_history", "question"], template=Template ) output_parser = StrOutputParser() model = ChatOpenAI( model_name=DEFAULT_MODEL, openai_api_key=APIKEY, openai_api_base=OPENAI_API_BASE, temperature=0.9, ) memory = ConversationSummaryMemory( llm=model, memory_key="chat_history", return_messages=True ) conversation_chain = ConversationalRetrievalChain.from_llm( llm=model, retriever=db.as_retriever(), memory=memory, combine_docs_chain_kwargs={'prompt': prompt}, verbose=False, ) ``` function calling : weather_function ```python class WeatherSearch(BaseModel): """Call this with an airport code to get the weather at that airport""" airport_code: str = Field(description="airport code to get weather for") weather_function = convert_pydantic_to_openai_function(WeatherSearch) ``` ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14988/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14987

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14987/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14987/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14987/events

https://github.com/langchain-ai/langchain/issues/14987

I_kwDOIPDwls56SQR1

Issue: Unable to return documents in my custom llm / agent executor implementation

open

NONE

### Issue you'd like to raise. I'm looking to use a HuggingFace pipeline using Mistral 7b. I am attempting to pass this into an AgentExectutor and use a retriever based tool. ```python from langchain.agents.agent_toolkits import create_retriever_tool from langchain_core.pydantic_v1 import BaseModel, Field class RetrieverInput(BaseModel): query: str = Field(description="query to look up in retriever") fantasy_football_tool = Tool( name="search_fantasy_football_articles", description="Searches and returns documents regarding fantasy football.", func=retriever.get_relevant_documents, # coroutine=retriever.aget_relevant_documents, args_schema=RetrieverInput, ) fantasy_football_tool.run("how is trevor lawrence doing?") [Document(page_content='Trevor Lawrence\n\nStill in concussion protocol Wednesday\n\nC.J. Stroud', metadata={'source': 'https://www.fantasypros.com/2023/11/rival-fantasy-nfl-week-10/'}), Document(page_content='Trevor Lawrence\n\nStill in concussion protocol Wednesday\n\nC.J. Stroud', metadata={'source': 'https://www.fantasypros.com/2023/11/nfl-week-10-sleeper-picks-player-predictions-2023/'}), ``` This shows that my tool is working as expected. Now to construct the the agent. ```oython prompt_template = """ ### [INST] Assistant is a large language model trained by Mistral. Assistant is designed to be able to assist with a wide range of tasks, from answering simple questions to providing in-depth explanations and discussions on a wide range of topics. As a language model, Assistant is able to generate human-like text based on the input it receives, allowing it to engage in natural-sounding conversations and provide responses that are coherent and relevant to the topic at hand. Assistant is constantly learning and improving, and its capabilities are constantly evolving. It is able to process and understand large amounts of text, and can use this knowledge to provide accurate and informative responses to a wide range of questions. Additionally, Assistant is able to generate its own text based on the input it receives, allowing it to engage in discussions and provide explanations and descriptions on a wide range of topics. Overall, Assistant is a powerful tool that can help with a wide range of tasks and provide valuable insights and information on a wide range of topics. Whether you need help with a specific question or just want to have a conversation about a particular topic, Assistant is here to assist. Context: ------ Assistant has access to the following tools: {tools} To use a tool, please use the following format: ``` Thought: Do I need to use a tool? Yes Action: the action to take, should be one of [{tool_names}] Action Input: the input to the action Observation: the result of the action ``` When you have a response to say to the Human, or if you do not need to use a tool, you MUST use the format: ``` Thought: Do I need to use a tool? No Final Answer: [your response here] ``` Begin! Previous conversation history: {chat_history} New input: {input} Current Scratchpad: {agent_scratchpad} [/INST] """ # Create prompt from prompt template prompt = PromptTemplate( input_variables=['agent_scratchpad', 'chat_history', 'input', 'tool_names', 'tools'], template=prompt_template, ) prompt = prompt.partial( tools=render_text_description(tools), tool_names=", ".join([t.name for t in tools]), ) # Create llm chain # This is a hugging face pipeline. llm_chain = LLMChain(llm=mistral_llm, prompt=prompt) from langchain.agents.conversational.output_parser import ConvoOutputParser from langchain.output_parsers.json import parse_json_markdown from langchain_core.exceptions import OutputParserException class CustomOutputParser(ConvoOutputParser): def parse(self, text: str) -> Union[AgentAction, AgentFinish]: """Attempts to parse the given text into an AgentAction or AgentFinish. Raises: OutputParserException if parsing fails. """ try: # If the response contains an 'action' and 'action_input' print(text) if "Action" in text or "Action Input" in text: # If the action indicates a final answer, return an AgentFinish if "Final Answer" in text: return AgentFinish({"output": text.split('Final Answer:')[1]}, text) else: # Otherwise, return an AgentAction with the specified action and # input return AgentAction(action, action_input, text) else: # If the necessary keys aren't present in the response, raise an # exception raise OutputParserException( f"Missing 'action' or 'action_input' in LLM output: {text}" ) except Exception as e: # If any other exception is raised during parsing, also raise an # OutputParserException raise OutputParserException(f"Could not parse LLM output: {text}") from e output_parser = CustomOutputParser() # Create an agent with your LLMChain agent = ConversationalAgent(llm_chain=llm_chain, output_parser=output_parser) memory = ConversationBufferMemory(memory_key="chat_history") agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True, memory=memory) ``` I've tested my `agent_executor` using the same question and get this: ```python Thought: Do I need to use a tool? Yes Action: search_fantasy_football_articles Action Input: "trevor lawrence" Observation: The search returned several articles discussing Trevor Lawrence's performance in fantasy football this week. Final Answer: According to the articles I found, Trevor Lawrence had a strong performance in fantasy football this week. ``` So it seems like it is pinging the tool but its not actually grabbing or using the documents. Any ideas on what I need to change? ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14987/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14986

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14986/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14986/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14986/events

https://github.com/langchain-ai/langchain/issues/14986

I_kwDOIPDwls56SHjN

Issue: Obtain the content output by AsyncCallbackHandler on_llm_new_token and send it to the front end to find that the newline character is missing.

closed

NONE

### Issue you'd like to raise. I used AsyncCallbackHandler for callback. When I pushed the content to the front end through on_llm_new_token, I found that the markdown code block was missing a newline character, which caused the front end to be unable to render the markdown format normally. However, when I retrieved the final response and returned the overall answer content, I found that this newline character existed. <img width="155" alt="image" src="https://github.com/langchain-ai/langchain/assets/14210962/214581c0-f427-4e83-b06e-f1ded11efe20"> I want to ask for help, how should I solve the problem I am currently facing? ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14986/timeline

completed

https://api.github.com/repos/langchain-ai/langchain/issues/14985

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14985/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14985/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14985/events

https://github.com/langchain-ai/langchain/pull/14985

PR_kwDOIPDwls5ihqTc

community[patch]: JaguarHttpClient conditional import

closed

CONTRIBUTOR

- **Description:** Fixed jaguar.py to import JaguarHttpClient with try and catch - **Issue:** the issue # Unable to use the JaguarHttpClient at run time - **Dependencies:** It requires "pip install -U jaguardb-http-client" - **Twitter handle:** workbot

https://api.github.com/repos/langchain-ai/langchain/issues/14985/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14984

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14984/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14984/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14984/events

https://github.com/langchain-ai/langchain/pull/14984

PR_kwDOIPDwls5ihoJI

Implement streaming for xml output parser

closed

COLLABORATOR

<!-- Thank you for contributing to LangChain! Please title your PR "<package>: <description>", where <package> is whichever of langchain, community, core, experimental, etc. is being modified. Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes if applicable, - **Dependencies:** any dependencies required for this change, - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` from the root of the package you've modified to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/docs/integrations` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. -->

https://api.github.com/repos/langchain-ai/langchain/issues/14984/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14983

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14983/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14983/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14983/events

https://github.com/langchain-ai/langchain/pull/14983

PR_kwDOIPDwls5ihmRd

Fixed jaguar.py to import JaguarHttpClient with try and catch

closed

CONTRIBUTOR

- **Description:** Fixed jaguar.py to import JaguarHttpClient with try and catch - **Issue:** the issue # Unable to use the JaguarHttpClient at run time - **Dependencies:** It requires "pip install -U jaguardb-http-client" - **Twitter handle:** workbot

https://api.github.com/repos/langchain-ai/langchain/issues/14983/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14982

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14982/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14982/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14982/events

https://github.com/langchain-ai/langchain/pull/14982

PR_kwDOIPDwls5ihjVm

community[patch]: Fix typo in class Docstring (#14982)

closed

CONTRIBUTOR

- **Description:** Fix typo in class Docstring to replace AZURE_OPENAI_API_ENDPOINT by AZURE_OPENAI_ENDPOINT - **Issue:** the issue #14901 - **Dependencies:** NA - **Twitter handle:**

https://api.github.com/repos/langchain-ai/langchain/issues/14982/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14981

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14981/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14981/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14981/events

https://github.com/langchain-ai/langchain/pull/14981

PR_kwDOIPDwls5ihabo

Implement streaming for all list output parsers

closed

COLLABORATOR

<!-- Thank you for contributing to LangChain! Please title your PR "<package>: <description>", where <package> is whichever of langchain, community, core, experimental, etc. is being modified. Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes if applicable, - **Dependencies:** any dependencies required for this change, - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` from the root of the package you've modified to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/docs/integrations` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. -->

https://api.github.com/repos/langchain-ai/langchain/issues/14981/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14980

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14980/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14980/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14980/events

https://github.com/langchain-ai/langchain/issues/14980

I_kwDOIPDwls56RtP9

ChatOllama stream method raises warn_deprecated NotImplementedError

open

NONE

### System Info langchain version: v0.0.352 python version: 3.11 Hi there! After that PR https://github.com/langchain-ai/langchain/pull/14713 was merged, I started getting errors in stream() method: ``` File .../lib/python3.11/site-packages/langchain_core/_api/deprecation.py:295, in warn_deprecated(since, message, name, alternative, pending, obj_type, addendum, removal) 293 if not removal: 294 removal = f"in {removal}" if removal else "within ?? minor releases" --> 295 raise NotImplementedError( 296 f"Need to determine which default deprecation schedule to use. " 297 f"{removal}" 298 ) 299 else: 300 removal = f"in {removal}" NotImplementedError: Need to determine which default deprecation schedule to use. within ?? minor releases ``` I guess this decorator must have a `pending=True` argument. ### Who can help? @hwchase17 @agola11 ### Information - [ ] The official example notebooks/scripts - [X] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [ ] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction ``` from langchain.chat_models import ChatOllama llm = ChatOllama( model="openchat:7b-v3.5-1210-q4_K_M", ) for chunk in llm.stream("Where were the Olympics held?"): print(chunk, end="", flush=True) ``` ### Expected behavior successful streaming output from llm

https://api.github.com/repos/langchain-ai/langchain/issues/14980/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14979

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14979/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14979/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14979/events

https://github.com/langchain-ai/langchain/issues/14979

I_kwDOIPDwls56Rsha

Ignoring Specific Paragraphs or Divs in Web Scraping with LangChain

closed

NONE

### Feature request ## Context: I am currently developing a custom scraper using the LangChain tools, following the provided documentation. The core functionality involves extracting paragraphs from a list of URLs using the AsyncHtmlLoader and the Beautiful Soup transformer: loader = AsyncHtmlLoader(urls) docs = loader.load() docs_transformed = self.bs_transformer.transform_documents(docs, tags_to_extract=["p"]) return docs_transformed ## Problem: The code successfully extracts all paragraphs from the provided URLs. However, in the case of web pages like https://www.aha.org/news/chairpersons-file/2023-12-18-chair-file-leadership-dialogue-reflecting-whats-next-health-care-joanne-conroy-md-dartmouth, there is a recurring issue. At the end of each blog or news article, there is a disclaimer message paragraph: "Noncommercial use of original content on www.aha.org is granted to AHA Institutional Members, their employees and State, Regional and Metro Hospital Associations unless otherwise indicated. AHA does not claim ownership of any content, including content incorporated by permission into AHA produced materials, created by any third party and cannot grant permission to use, distribute or otherwise reproduce such third party content. To request permission to reproduce AHA content, please [click here](https://askrc.libraryresearch.info/reft100.aspx?key=ExtPerm). " ## Proposed Solution: To address this, I explored options and realized that excluding specific parts of the HTML could be a viable solution. Typically, using Beautiful Soup, I can delete specific paragraphs within a div by targeting the class parameter, as demonstrated here: soup.find('div', class_='aha-footer') ## Issue with LangChain Implementation: Upon inspecting the beautiful_soup_transformer.py in the LangChain repository, particularly the remove_unwanted_tags method, I observed that it is currently implemented to remove unwanted tags in a general sense: soup = BeautifulSoup(html_content, "html.parser") for tag in unwanted_tags: for element in soup.find_all(tag): element.decompose() return str(soup) This implementation makes it impossible to selectively eliminate specific divs from the HTML. ## Request for Guidance: I seek guidance on how to ignore specific paragraphs or divs during web scraping with LangChain, particularly to exclude the recurring disclaimer paragraph mentioned above. I would appreciate any recommendations on the recommended approach or if there are plans to enhance the beautiful_soup_transformer.py to accommodate more granular exclusion of HTML elements. ### Motivation I am performing web scrapping over this specific web page: https://www.aha.org/news/chairpersons-file/2023-12-18-chair-file-leadership-dialogue-reflecting-whats-next-health-care-joanne-conroy-md-dartmouth I am extracting all the paragraphs , but at the end of all blogs, news there is a disclaimer message paragraph: Noncommercial use of original content on www.aha.org is granted to AHA Institutional Members, their employees and State, Regional and Metro Hospital Associations unless otherwise indicated. AHA does not claim ownership of any content, including content incorporated by permission into AHA produced materials, created by any third party and cannot grant permission to use, distribute or otherwise reproduce such third party content. To request permission to reproduce AHA content, please [click here](https://askrc.libraryresearch.info/reft100.aspx?key=ExtPerm). so I want to ignore that specific paragraph ### Your contribution not yet

https://api.github.com/repos/langchain-ai/langchain/issues/14979/timeline

completed

https://api.github.com/repos/langchain-ai/langchain/issues/14978

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14978/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14978/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14978/events

https://github.com/langchain-ai/langchain/pull/14978

PR_kwDOIPDwls5ihRu9

community[patch]: support momento vector index filter expressions

closed

CONTRIBUTOR

**Description** For the Momento Vector Index (MVI) vector store implementation, pass through `filter_expression` kwarg to the MVI client, if specified. This change will enable the MVI self query implementation in a future PR. Also fixes some integration tests.

https://api.github.com/repos/langchain-ai/langchain/issues/14978/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14976

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14976/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14976/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14976/events

https://github.com/langchain-ai/langchain/pull/14976

PR_kwDOIPDwls5ihIMF

API Ref navbar update

open

COLLABORATOR

Now the navbar items are too long and are not visible.  I'm trying to cut the top-level part of the namespace in items.

https://api.github.com/repos/langchain-ai/langchain/issues/14976/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14975

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14975/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14975/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14975/events

https://github.com/langchain-ai/langchain/issues/14975

I_kwDOIPDwls56Rane

Issue: Requesting Feedback on Integrating Gretel for Synthetic Tabular Generation

open

NONE

### Issue you'd like to raise. Proposing updates to the SyntheticDataGenerator interface to create a cleaner foundation for building community integrations for synthetic tabular models like [Gretel](https://gretel.ai/tabular-llm) [[docs](https://docs.gretel.ai/reference/tabular-llm)] ### Suggestion: # Existing Interface The current `SyntheticDataGenerator` interface requires: ```python def generate( subject: str, runs: int, extra: Optional[str] = None ) -> List[str] ``` Where: * `subject`: Subject the synthetic data is about * `runs`: Number of times to generate the data * `extra`: Extra instructions for steering ## Proposed Update I propose changing this to the following: ```python def generate( prompt: str, num_records: int, optional_dataset: Optional[str, Path, DataFrame] ) -> List[str] ``` ## Where: * `prompt`: User prompt to create synthetic data * `num_records`: Number of rows to generate * `optional_dataset`: Dataset to edit/augment I believe this creates a cleaner interface for synthetic tabular data flows, by combining the `subject` and `extra` parameters into a single field, and allowing the user to clearly specify the number of results they want `num_records`, vs the number of `runs` of the LLM which could generate more than one record each time. The `optional_dataset` arg lets the user prompt the model with a dataset to edit or augment with new synthetic data. # Requesting Feedback I would appreciate any thoughts on this proposed update, and happy to open a PR! Before I get started, please let me know: * If you see any issues with changing the interface * If an alternative integration approach would be better * Any other API details to consider https://python.langchain.com/docs/use_cases/data_generation My goal is have an intuitive integration for Gretel and future synthetic data models

https://api.github.com/repos/langchain-ai/langchain/issues/14975/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14974

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14974/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14974/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14974/events

https://github.com/langchain-ai/langchain/issues/14974

I_kwDOIPDwls56RUsj

LLMs start replying in other languages

open

NONE

### System Info Currently, I am using OpenAI LLM and Gemini Pro all being used my LangChain. I am also using Google's embedding-001 model and Cohere base model (tested each embedding and both either reply back in english first then another language or straight to another language). Here is my prompt template: <code> def doc_question_prompt_template(): template = """ You are a helpful assistant that has the ability to answer all users questions to the best of your ability. Your answers should come from the context you are provided. Provide an answer with detail and not short answers. Your only response should be in the English langeuage. Context: {context} User: {question} """ return PromptTemplate( input_variables=["question"], template=template ) def doc_question_command(body, conversation_contexts): llmlibrary = LLMLibrary() channel_id = body['channel_id'] user_id = body['user_id'] context_key = f"{channel_id}-{user_id}" prompt = ChatPromptTemplate.doc_question_prompt_template() if context_key not in conversation_contexts: conversation_contexts[context_key] = { "memory": ConversationBufferMemory(memory_key="chat_history", output_key="answer", return_messages=True, max_token_limit=1024), "history": "", } user_memory = conversation_contexts[context_key]["memory"] question = body['text'] conversation = llmlibrary.doc_question(user_memory, prompt, question) #print(f"Conversation: {conversation}") return question, conversation def doc_question(self, user_memory, prompt, question) llm = ChatGoogleGenerativeAI(model="gemini-pro",temperature=0.0, convert_system_message_to_human=True) vectordb = self.vectorstore.get_vectordb() print(f"Vector DB: {vectordb}\n") retriever = vectordb.as_retriever( search_type="similarity_score_threshold", search_kwargs={'score_threshold': 0.8} ) docs = retriever.get_relevant_documents(question) print(f"Docs: {docs}\n") print(f"Initiating chat conversation memory\n") #print(f"Conversation Memory: {memory}\n") conversation_chain= ConversationalRetrievalChain.from_llm( llm, retriever=retriever, memory=user_memory, combine_docs_chain_kwargs={'prompt': prompt}, return_source_documents=True, verbose=False, ) #print(f"Conversation chain: {conversation_chain}\n") return conversation_chain @app.command("/doc_question") def handle_doc_question_command(ack, body, say): # Acknowledge the command request ack() print(body) say(f"🤨 {body['text']}") question, conversation = ChatHandler.doc_question_command(body, conversation_contexts) response = conversation({'question': question}) print(f"(INFO) Doc Question Response: {response} {time.time()}") print(f"(INFO) Doc Question Response answer: {response['answer']} {time.time()}") say(f"🤖 {response['answer']}") </code> Logs: [output.txt](https://github.com/langchain-ai/langchain/files/13732990/output.txt)  ### Who can help? _No response_ ### Information - [ ] The official example notebooks/scripts - [ ] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [ ] Embedding Models - [X] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction 1. user sends a message through Slack 2. The message is received by @app.command("/doc_question) 3. ChatHandler.doc_question_command gets called passing the body and conversation_contexts 4. doc_question_command gets information about the message that was sent and gets the doc_question_prompt_template from ChatPromptTemplate module 5. conversation_contexts gets a context key of memory and history 6. llmlibrary.doc question is then called passing user_member, prompt, question 7. the doc_question function uses the ChatGoogleGenerativeAI module and gets the vectordb which is Pinecone 8. uses the ConversationRetrievalChain.from_llm and passes it back to the handler and the handler passes question and conversation back to @app.command("/doc_question") 9. The question is then submitted to the LLM and the response is spat out within Slack (sometimes english, Spanish, or other) ### Expected behavior Only reply in english

https://api.github.com/repos/langchain-ai/langchain/issues/14974/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14973

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14973/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14973/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14973/events

https://github.com/langchain-ai/langchain/pull/14973

PR_kwDOIPDwls5ig25f

community: add args_schema to GmailSendMessage

closed

CONTRIBUTOR

- **Description:** `tools.gmail.send_message` implements a `SendMessageSchema` that is not used anywhere. `GmailSendMessage` also does not have an `args_schema` attribute (this led to issues when invoking the tool with an OpenAI functions agent, at least for me). Here we add the missing attribute and a minimal test for the tool. - **Issue:** N/A - **Dependencies:** N/A - **Twitter handle:** N/A

https://api.github.com/repos/langchain-ai/langchain/issues/14973/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14972

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14972/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14972/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14972/events

https://github.com/langchain-ai/langchain/pull/14972

PR_kwDOIPDwls5ig0qb

ci test - do not merge

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14972/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14971

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14971/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14971/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14971/events

https://github.com/langchain-ai/langchain/pull/14971

PR_kwDOIPDwls5igk7z

WIP Runnable Chain

closed

COLLABORATOR

<!-- Thank you for contributing to LangChain! Please title your PR "<package>: <description>", where <package> is whichever of langchain, community, core, experimental, etc. is being modified. Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes if applicable, - **Dependencies:** any dependencies required for this change, - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` from the root of the package you've modified to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/docs/integrations` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. -->

https://api.github.com/repos/langchain-ai/langchain/issues/14971/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14970

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14970/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14970/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14970/events

https://github.com/langchain-ai/langchain/pull/14970

PR_kwDOIPDwls5igR2B

Vectara summarization

closed

COLLABORATOR

Description: Adding Summarization to Vectara, to reflect it provides not only vector-store type functionality but also can return a summary. Also added: MMR capability (in the Vectara platform side) Updated templates Updated documentation and IPYNB examples Tag maintainer: @baskaryan Twitter handle: @ofermend

https://api.github.com/repos/langchain-ai/langchain/issues/14970/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14969

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14969/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14969/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14969/events

https://github.com/langchain-ai/langchain/pull/14969

PR_kwDOIPDwls5igPKS

Optionally limit the Current Buffer loaded in ConversationSummaryBufferMemory

open

CONTRIBUTOR

- **Description:** - `ConversationSummaryBufferMemory` does not have a way to limit the loaded messages/buffer from `ChatHistory`. While it summarizes/prunes the previous buffer when the buffer loaded from `ChatHistory` exceeds the token threshold but when `ConversationSummaryBufferMemory` loads the memory variables, it reloads all of messages which goes along with `moving_summary_buffer`. Their needs to be a way to OPTIONALLY limit the Window of current buffer loaded from ChatHistory which will go with the moving_summary_buffer in the final prompt otherwise the memory variables will just keep on increasing like simple `ConversationBufferMemory`. - **Issue:** #14822 - **Tag maintainer:** @baskaryan, @eyurtsev, @hwchase17.

https://api.github.com/repos/langchain-ai/langchain/issues/14969/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14967

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14967/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14967/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14967/events

https://github.com/langchain-ai/langchain/pull/14967

PR_kwDOIPDwls5ifgUV

core(minor): Allow explicit types for ChatMessageHistory adds

closed

CONTRIBUTOR

<!-- Thank you for contributing to LangChain! Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes (if applicable), - **Dependencies:** any dependencies required for this change, - **Tag maintainer:** for a quicker response, tag the relevant maintainer (see below), - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/extras` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. --> ## Description Changes the behavior of `add_user_message` and `add_ai_message` to allow for messages of those types to be passed in. Currently, if you want to use the `add_user_message` or `add_ai_message` methods, you have to pass in a string. For `add_message` on `ChatMessageHistory`, however, you have to pass a `BaseMessage`. This behavior seems a bit inconsistent. Personally, I'd love to be able to be explicit that I want to `add_user_message` and pass in a `HumanMessage` without having to grab the `content` attribute. This PR allows `add_user_message` to accept `HumanMessage`s or `str`s and `add_ai_message` to accept `AIMessage`s or `str`s to add that functionality and ensure backwards compatibility. ## Issue * None ## Dependencies * None ## Tag maintainer @hinthornw @baskaryan ## Note `make test` results in `make: *** No rule to make target 'test'. Stop.`

https://api.github.com/repos/langchain-ai/langchain/issues/14967/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14966

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14966/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14966/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14966/events

https://github.com/langchain-ai/langchain/pull/14966

PR_kwDOIPDwls5ifevu

packaging: fix format for python version

closed

NONE

Hello, could you adjust the version format? I am running into issues where it's not understood by some tools. The python version format is only major.minor. It's one of the edge cases of python packaging. You can see examples in setuptools, both the required python version for the package and the required python version for dependencies. It's not documented well but the format is only meant to be two digits. https://setuptools.pypa.io/en/latest/userguide/declarative_config.html https://setuptools.pypa.io/en/latest/userguide/pyproject_config.html ``` # setup.cfg [options] zip_safe = False include_package_data = True packages = find: python_requires = >=3.7 install_requires = requests importlib-metadata; python_version<"3.8" ```

https://api.github.com/repos/langchain-ai/langchain/issues/14966/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14964

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14964/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14964/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14964/events

https://github.com/langchain-ai/langchain/pull/14964

PR_kwDOIPDwls5ifb1V

feat: add streaming param to ChatMistralAI constructor

closed

CONTRIBUTOR

- **Description:** A new boolean parameter `streaming` has been added to the constructor of the `ChatMistralAI` class. This parameter allows users to specify whether they wish to use the streaming functionality at the time of instantiating a `ChatMistralAI` object. - **Tag maintainer:** @baskaryan - **Twitter handle:** @LMS_David_RS

https://api.github.com/repos/langchain-ai/langchain/issues/14964/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14963

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14963/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14963/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14963/events

https://github.com/langchain-ai/langchain/pull/14963

PR_kwDOIPDwls5ifZ50

infra: pr template update

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14963/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14961

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14961/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14961/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14961/events

https://github.com/langchain-ai/langchain/pull/14961

PR_kwDOIPDwls5ie4U-

langchain[patch]: Release 0.0.352

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14961/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14960

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14960/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14960/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14960/events

https://github.com/langchain-ai/langchain/pull/14960

PR_kwDOIPDwls5ie2kv

community[patch]: Release 0.0.5

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14960/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14959

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14959/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14959/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14959/events

https://github.com/langchain-ai/langchain/pull/14959

PR_kwDOIPDwls5ie0Rq

core[patch]: 0.1.2

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14959/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14958

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14958/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14958/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14958/events

https://github.com/langchain-ai/langchain/pull/14958

PR_kwDOIPDwls5ieqEf

Dmaturana/randomlengthfewshotselector

closed

CONTRIBUTOR

- **Description:** The length example selector always retrieves the same examples, going from top to bottom. In the case you have a lot of examples, there will not be any variation. I just copied and modified this length selector to do a random length base examples selector, so for each prompt, the examples selected will be random - **Dependencies:** Added the random import, not dependencies changed - **Tag maintainer:** for a quicker response, tag the relevant maintainer (see below),

https://api.github.com/repos/langchain-ai/langchain/issues/14958/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14957

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14957/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14957/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14957/events

https://github.com/langchain-ai/langchain/issues/14957

I_kwDOIPDwls56OjPN

Issue with DynamoDBChatMessageHistory Not Respecting max_token_limit in ConversationTokenBufferMemory

open

NONE

### System Info Langchain Version: langchain 0.0.350, langchain-community 0.0.3, langchain-core 0.1.1 Python Version: 3.10.6 Operating System: macOs Additional Libraries: boto 2.49.0, boto3 1.34.1 ### Who can help? _No response_ ### Information - [ ] The official example notebooks/scripts - [X] My own modified scripts ### Related Components - [ ] LLMs/Chat Models - [ ] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [X] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction Steps to Reproduce: - Create an instance of `DynamoDBChatMessageHistory` with a specified table name and session ID. - Initialize `ConversationTokenBufferMemory` with a `max_token_limit`. - Attach the memory to a `ConversationChain`. - Call predict on the `ConversationChain` with some input. Code sample: ``` import boto3 from langchain.llms import Bedrock from langchain.memory.chat_message_histories import DynamoDBChatMessageHistory from langchain.memory import ConversationTokenBufferMemory session = boto3.Session( aws_access_key_id=os.environ.get('AWS_ACCESS_KEY_ID'), aws_secret_access_key=os.environ.get('AWS_SECRET_ACCESS_KEY'), aws_session_token=os.environ.get('AWS_SESSION_TOKEN'), region_name='us-east-1' ) dynamodb = session.resource('dynamodb') chat_sessions_table = dynamodb.Table('SessionTable') boto3_bedrock = session.client(service_name="bedrock-runtime") max_tokens_to_sample = 100 temperature = 0 modelId = "anthropic.claude-instant-v1" top_k = 250 top_p = 0.999 model_kwargs = { "temperature": temperature, "max_tokens_to_sample": max_tokens_to_sample, "top_k": top_k, "top_p": top_p } llm = Bedrock( client=boto3_bedrock, model_id=modelId, region_name='us-east-1', model_kwargs=model_kwargs, streaming=True,callbacks=[StreamingStdOutCallbackHandler()] ) message_history = DynamoDBChatMessageHistory(table_name="SessionTable", session_id="10", boto3_session=session) memory = ConversationTokenBufferMemory( llm=llm, # Use the Bedrock instance max_token_limit=100, return_messages=True, chat_memory=message_history, ai_prefix="A", human_prefix="H" ) #add the memory to the Chain conversation = ConversationChain( llm=llm, verbose=True, memory=memory ) conversation.predict(input="Hello!") memory.load_memory_variables({}) ``` ### Expected behavior Expected Behavior: - The `DynamoDBChatMessageHistory` should respect the `max_token_limit` set in `ConversationTokenBufferMemory`, limiting the token count accordingly. Actual Behavior: The `DynamoDBChatMessageHistory` does not limit the token count as per the `max_token_limit` set in `ConversationTokenBufferMemory` and keeps saving all the items in memory on the DynamoDB table.

https://api.github.com/repos/langchain-ai/langchain/issues/14957/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14956

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14956/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14956/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14956/events

https://github.com/langchain-ai/langchain/issues/14956

I_kwDOIPDwls56OYg7

Feature Request - ChatGPT OpenAPI NLATool Authentication Implementation Logic

open

NONE

### Feature request **Issue Title:** Enhance NLATool Authentication in ChatGPT OpenAPI **Description:** I have identified a feature gap in the current implementation of ChatGPT OpenAPI when using NLATool as a proxy for authentication. The existing logic does not fully meet downstream requirements, and I intend to propose a modification to address this issue. **Proposed Modification:** I suggest adding a new attribute within the NLATool implementation. During initialization, this attribute should be passed to `NLAToolkit.from_llm_and_ai_plugin`. The subsequent call chain is as follows: `from_llm_and_spec -> _get_http_operation_tools -> NLATool.from_llm_and_method`. The responsibility of `NLATool.from_llm_and_method` is to construct the NLATool component, which includes an underlying `OpenAPIEndpointChain` base package. The challenge lies in the fact that the `OpenAPIEndpointChain` base package currently lacks support for authentication. To address this, it is essential to load the created attribute into the `OpenAPIEndpointChain`. During the execution of the `_call` method, the authentication logic should be executed. **Implementation Steps:** 1. Modify the initialization and execution code of the `OpenAPIEndpointChain` class to support authentication. 2. Ensure that the newly added attribute is properly integrated into the `OpenAPIEndpointChain` during its initialization. 3. Implement the authentication logic in the `_call` method of the `OpenAPIEndpointChain`. Thank you for your consideration. ### Motivation The current implementation of ChatGPT OpenAPI using NLATool as a proxy for authentication falls short of meeting downstream requirements. By enhancing the NLATool authentication logic, we aim to improve the overall functionality and responsiveness of the system, ensuring it aligns more effectively with user needs. This modification seeks to bridge the existing feature gap and enhance the usability and versatility of the ChatGPT OpenAPI. ### Your contribution **Expected Challenges:** While I have not yet started the implementation, I anticipate challenges during the process. One notable challenge is that the `langchain` module does not currently define the core for the authentication class. Consequently, addressing this issue may require changes across multiple modules. A pull request spanning multiple modules may encounter challenges during the review process. **Request for Feedback:** Before I commence with the implementation, I would appreciate your insights and guidance on the proposed modification. Your feedback on potential challenges and recommendations for an effective implementation would be invaluable.

https://api.github.com/repos/langchain-ai/langchain/issues/14956/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14954

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14954/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14954/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14954/events

https://github.com/langchain-ai/langchain/issues/14954

I_kwDOIPDwls56Nwov

getting the error "TypeError: Agent.plan() got multiple values for argument 'intermediate_steps'" with agent_executor

open

NONE

### System Info langchain: 0.0.351 langchain-community: 0.0.4 langchain-core: 0.1.1 langchain-experimental: 0.0.47 python: 3.10.4 ### Who can help? @hwchase17 , @agola11 , @eyurtsev ### Information - [X] The official example notebooks/scripts - [X] My own modified scripts ### Related Components - [ ] LLMs/Chat Models - [ ] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [X] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [X] Agents / Agent Executors - [X] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction 1. Create an agent and which is working as expected. 2. create an agent_executor by using above agent. 3. If i try to use agent_execuor, getting error "TypeError: Agent.plan() got multiple values for argument 'intermediate_steps'" **Below is the code:** Create an agent:- ``` agent = initialize_agent(llm=llm, tools=tools, agent = AgentType.CONVERSATIONAL_REACT_DESCRIPTION, verbose=True, agent_kwargs=agent_kwargs, output_parser = output_parser, output_key = "result", handle_parsing_errors = True, max_iterations=3, early_stopping_method="generate", memory = memory, ) ``` Create an agent_executor:- ``` agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True, memory = memory, ) ``` calling the agent_executor `result = agent_executor.invoke({"input":"Tell me about yourself", "format_instructions": response_format})["output"]` Getting below error:- ``` Entering new AgentExecutor chain... > Entering new AgentExecutor chain... --------------------------------------------------------------------------- TypeError Traceback (most recent call last) Cell In[18], line 1 ----> 1 result = agent_executor.invoke({"input":"Tell me about yourself", "format_instructions": response_format})["output"] 2 print(f"result: {result}") TypeError: Agent.plan() got multiple values for argument 'intermediate_steps' ``` What I have observed from the above error is that the chain is executing multiple times, hence the 'Entering new AgentExecutor chain...' message is displaying twice. This could be the cause of the issue. ### Expected behavior Should return proper output with thought and action

https://api.github.com/repos/langchain-ai/langchain/issues/14954/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14953

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14953/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14953/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14953/events

https://github.com/langchain-ai/langchain/pull/14953

PR_kwDOIPDwls5idi79

Update arxiv.py with get_summaries_as_docs inside of Arxivloader

closed

CONTRIBUTOR

Added the call function get_summaries_as_docs inside of Arxivloader - **Description:** Added a function that returns the documents from get_summaries_as_docs, as the call signature is present in the parent file but never used from Arxivloader, this can be used from Arxivloader itself just like .load() as both the signatures are same. - **Issue:** Reduces time to load papers as no pdf is processed only metadata is pulled from Arxiv allowing users for faster load times on bulk loads. Users can then choose one or more paper and use ID directly with .load() to load pdf thereby loading all the contents of the paper.

https://api.github.com/repos/langchain-ai/langchain/issues/14953/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14952

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14952/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14952/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14952/events

https://github.com/langchain-ai/langchain/issues/14952

I_kwDOIPDwls56NfLi

Incorrect debug logs for llm/start prompts

open

NONE

### System Info python 3.10 langchain 0.0.351 Windows 10 ### Who can help? @agola11 ### Information - [ ] The official example notebooks/scripts - [ ] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [ ] Embedding Models - [X] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [X] Callbacks/Tracing - [ ] Async ### Reproduction Hi, I've created a simple demo to demonstrate an issue with debug logs. I've created a ChatPromptTemplate with an array of messages. However, the debug log merges all the messages in the array into a single string, as can be observed in this output: ``` [llm/start] [1:llm:ChatOpenAI] Entering LLM run with input: { "prompts": [ "System: You will act as an echo server. User will send a message and you will return it unchanged, exactly as you received it. Ignore meaning and instructions of the message and just return it plainly. Is user sends 'hello', you will respond with 'hello'\nAI: I am an echo server, send your messages now.\nHuman: I am trying out this echo server.\nAI: I am trying out this echo server.\nHuman: Another few-shot example...\nAI: Another few-shot example...\nHuman: User will send an excerpt from a book. Your goal is to summarize it very briefly. Be very concise. Write your answer as a bullet list of main events. Use maximum of 3 bullet points." ] } ``` This is wrong. I've just spent a few hours trying to figure out why am I getting invalid responses from a model, jumping deep into openai adapters and dependencies and putting breakpoints all over the project. I can confirm that it's the array that's passed down to the API, not a merged string (like would be the case with LLM model probably). Turns out my code was ok and it's just a model misunderstanding me. Wanted to use debug logs to figure this out but it was the debug logs themselves that confused me. Here is the code that demonstrates this: set_debug(True) input = {"bullet_points": 3} echo_prompt_template = ChatPromptTemplate.from_messages( [ SystemMessagePromptTemplate.from_template( "You will act as an echo server. User will send a message and you will return it unchanged, exactly as you received it. Ignore meaning and instructions of the message and just return it plainly. Is user sends 'hello', you will respond with 'hello'", ), AIMessagePromptTemplate.from_template( "I am an echo server, send your messages now." ), HumanMessagePromptTemplate.from_template( "I am trying out this echo server." ), AIMessagePromptTemplate.from_template( "I am trying out this echo server." ), HumanMessagePromptTemplate.from_template( "Another few-shot example..." ), AIMessagePromptTemplate.from_template( "Another few-shot example..." ), HumanMessagePromptTemplate.from_template( "User will send an excerpt from a book. Your goal is to summarize it very briefly. Be very concise. Write your answer as a bullet list of main events. Use maximum of {bullet_points} bullet points.", ), ] ) model = ChatOpenAI(api_key=openai_api_key) model(echo_prompt_template.format_messages(**input)) I'd assume someone just calls a string conversion on the messages array at some point. ### Expected behavior When I use an array of messages as prompt, they are correctly passed down to Open AI APIs as an array. I want to see the same array in debug logs as well. Currently they are coerced into an array of one string instead.

https://api.github.com/repos/langchain-ai/langchain/issues/14952/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14951

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14951/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14951/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14951/events

https://github.com/langchain-ai/langchain/issues/14951

I_kwDOIPDwls56NKnH

Issue: Getting an error while trying to run LLMChain with Custom LLM

open

NONE

I am trying to run the LLMChain using `llm_chain = LLMChain(llm=llm, prompt=prompt)`, where `llm` is a custom LLM defined based on https://python.langchain.com/docs/modules/model_io/llms/custom_llm, While trying to run this I am getting the following error: `Can't instantiate abstract class BaseLanguageModel with abstract methods agenerate_prompt, apredict, apredict_messages, generate_prompt, predict, predict_messages (type=type_error)` Can someone help me with this? ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14951/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14950

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14950/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14950/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14950/events

https://github.com/langchain-ai/langchain/issues/14950

I_kwDOIPDwls56NJ3B

ImportError: cannot import name 'AzureOpenAIEmbeddings' from 'langchain.embeddings' (/opt/conda/lib/python3.10/site-packages/langchain/embeddings/__init__.py)

open

NONE

### System Info unable to resolve below issue  ### Who can help? _No response_ ### Information - [X] The official example notebooks/scripts - [ ] My own modified scripts ### Related Components - [ ] LLMs/Chat Models - [X] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction  ### Expected behavior ImportError: cannot import name 'AzureOpenAIEmbeddings' from 'langchain.embeddings' (/opt/conda/lib/python3.10/site-packages/langchain/embeddings/__init__.py)

https://api.github.com/repos/langchain-ai/langchain/issues/14950/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14949

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14949/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14949/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14949/events

https://github.com/langchain-ai/langchain/pull/14949

PR_kwDOIPDwls5ic8iQ

Add support Vertex AI Gemini uses a public image URL

closed

CONTRIBUTOR

## What Since `langchain_google_genai.ChatGoogleGenerativeAI` supported A public image URL, we add to support it in `langchain.chat_models.ChatVertexAI` as well. ### Example ```py from langchain.chat_models.vertexai import ChatVertexAI from langchain_core.messages import HumanMessage llm = ChatVertexAI(model_name="gemini-pro-vision") image_message = { "type": "image_url", "image_url": { "url": "https://python.langchain.com/assets/images/cell-18-output-1-0c7fb8b94ff032d51bfe1880d8370104.png", }, } text_message = { "type": "text", "text": "What is shown in this image?", } message = HumanMessage(content=[text_message, image_message]) output = llm([message]) print(output.content) ``` ## Refs - https://python.langchain.com/docs/integrations/llms/google_vertex_ai_palm - https://python.langchain.com/docs/integrations/chat/google_generative_ai

https://api.github.com/repos/langchain-ai/langchain/issues/14949/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14948

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14948/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14948/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14948/events

https://github.com/langchain-ai/langchain/issues/14948

I_kwDOIPDwls56MypP

The calculated score is wrong when using DistanceStrategy.MAX_INNER_PRODUCT (Faiss)

open

NONE

### System Info langchain: 0.0.338 python: 3.9 ### Who can help? @hwchase17 @eyurtsev ### Information - [ ] The official example notebooks/scripts - [X] My own modified scripts ### Related Components - [ ] LLMs/Chat Models - [X] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [X] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction In the following code: ``` python if self.distance_strategy == DistanceStrategy.MAX_INNER_PRODUCT: return self._max_inner_product_relevance_score_fn elif self.distance_strategy == DistanceStrategy.EUCLIDEAN_DISTANCE: # Default behavior is to use Euclidean distance for relevancy return self._euclidean_relevance_score_fn elif self.distance_strategy == DistanceStrategy.COSINE: return self._cosine_relevance_score_fn ``` When I use MAX_INNER_PRODUCT, the score calculation method is `_max_inner_product_relevance_score_fn`: ``` python def _max_inner_product_relevance_score_fn(distance: float) -> float: """Normalize the distance to a score on a scale of [0, 1].""" if distance > 0: return 1.0 - distance return -1.0 * distance ``` However, if I use MAX_INNER_PRODUCT, the index must be FlatIP: ``` python if distance_strategy == DistanceStrategy.MAX_INNER_PRODUCT: index = faiss.IndexFlatIP(len(embeddings[0])) else: # Default to L2, currently other metric types not initialized. index = faiss.IndexFlatL2(len(embeddings[0])) ``` Thus, the distance represents the cosine similarity, meaning the distance should be equivalent to similarity. However, in the method `_max_inner_product_relevance_score_fn`, a larger distance results in a lower score. Is this a bug? ### Expected behavior I think the distance sholud be equivalent to similarity。

https://api.github.com/repos/langchain-ai/langchain/issues/14948/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14947

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14947/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14947/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14947/events

https://github.com/langchain-ai/langchain/issues/14947

I_kwDOIPDwls56MyGb

Issue: Getting 'An output parsing error occurred' error even after passing 'handle_parsing_errors=True' to the agent

open

NONE

### Issue you'd like to raise. I am using langchain with gpt 4 model. Im using the create pandas dataframe agent for my use case. For 60 % of the time i run the code, im getting below error- An output parsing error occurred. In order to pass this error back to the agent and have it try again, pass `handle_parsing_errors=True` to the AgentExecutor. This is the error: Could not parse LLM output: `Thought: To answer the question about what the "EffectiveDate" column represents, I need to use common sense based on the column name and the data provided. For the "exists" part, I need to check if there are any standardization issues in the "EffectiveDate" column. I will look at the data provided and think about the possible issues listed. Now i have already passed the argument- `handle_parsing_errors=True' while creating the agent, but still it gives me the above error, suggesting me to pass this argument. I have also tried giving other values to the handle_parsing_errors like - passing custom error message or function, still im stuck with this error most of the time. ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14947/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14946

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14946/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14946/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14946/events

https://github.com/langchain-ai/langchain/issues/14946

I_kwDOIPDwls56MVge

openai migrate

open

NONE

### System Info loader1 = CSVLoader(file_path='/home/calvin/下载/test.csv') Doc = loader1.load() text_splitter = CharacterTextSplitter(chunk_size=100,chunk_overlap=0) texts = text_splitter.split_documents(Doc) embeddings = OpenAIEmbeddings() db = Chroma.from_documents(texts, embeddings) retriever = db.as_retriever() qa = RetrievalQA.from_chain_type(llm=OpenAI(mode="gpt-3.5-turbo"), chain_type="stuff", retriever=retriever) query = "1501475820" print(qa.run(query)) I run this code,but i can not use chat-gpt-3.5-turbo，so i try to opanAI MIGRATE，but i exit it ,then i found  then always tell me  @hwcha ### Who can help? from langchain.chains import RetrievalQA ### Information - [X] The official example notebooks/scripts - [X] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [ ] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction loader1 = CSVLoader(file_path='/home/calvin/下载/test.csv') Doc = loader1.load() text_splitter = CharacterTextSplitter(chunk_size=100,chunk_overlap=0) texts = text_splitter.split_documents(Doc) embeddings = OpenAIEmbeddings() db = Chroma.from_documents(texts, embeddings) retriever = db.as_retriever() qa = RetrievalQA.from_chain_type(llm=OpenAI(mode="gpt-3.5-turbo"), chain_type="stuff", retriever=retriever) query = "1501475820" print(qa.run(query)) ### Expected behavior i want use gpt-3.5-turbo to query

https://api.github.com/repos/langchain-ai/langchain/issues/14946/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14945

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14945/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14945/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14945/events

https://github.com/langchain-ai/langchain/issues/14945

I_kwDOIPDwls56Lofa

Issue: how to deploy gpt-4-turbo through langchain

closed

NONE

### Issue you'd like to raise. Hi, I can't find from how to deploy gpt-4-turbo in langchain. Could anyone please tell me throuhg which module gpt-4-turbo can be deployed? Seems that the langchain.llm has already been removed from new version of langchain. ### Suggestion: _No response_

https://api.github.com/repos/langchain-ai/langchain/issues/14945/timeline

completed

https://api.github.com/repos/langchain-ai/langchain/issues/14944

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14944/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14944/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14944/events

https://github.com/langchain-ai/langchain/issues/14944

I_kwDOIPDwls56LlSU

ImportError: cannot import name 'LLMContentHandler' from 'langchain.llms.sagemaker_endpoint' occurring with 0.0.351

open

NONE

### System Info Langchain == 0.0.351 Python == 3.10.6 Running in AWS sagemaker notebook, issue occurred on multiple kernels. Code worked perfectly yesterday, error occurred upon starting up this morning (12/19/23) Code worked again upon reverting to 0.0.349 ### Who can help? _No response_ ### Information - [ ] The official example notebooks/scripts - [X] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [ ] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [ ] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction pip install -U langchain from langchain.llms.sagemaker_endpoint import LLMContentHandler ### Expected behavior Expected behavior is that the import works

https://api.github.com/repos/langchain-ai/langchain/issues/14944/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14943

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14943/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14943/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14943/events

https://github.com/langchain-ai/langchain/issues/14943

I_kwDOIPDwls56LUQu

How does AgentExecutor make LLM on_llm_new_token most streaming output instead of AgentExecutorIterator?

open

NONE

How does AgentExecutor make LLM on_llm_new_token most streaming output instead of AgentExecutorIterator? The current effect is that it will stream out AgentExecutorIterator Desired effect: streams LLM on_llm_new_token

https://api.github.com/repos/langchain-ai/langchain/issues/14943/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14942

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14942/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14942/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14942/events

https://github.com/langchain-ai/langchain/pull/14942

PR_kwDOIPDwls5ibh7M

update notebook documentation for custom tool

closed

CONTRIBUTOR

- **Description:** Documentation update. The custom tool notebook documentation is updated to revome the warning caused by directly instantiating of the LLMMathChain with an llm which is is deprecated. The from_llm class method is used instead. LLM output results gets updated as well. - **Issue:** no applicable - **Dependencies:** No dependencies - **Tag maintainer:** @baskaryan - **Twitter handle:** @ybouakkaz

https://api.github.com/repos/langchain-ai/langchain/issues/14942/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14941

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14941/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14941/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14941/events

https://github.com/langchain-ai/langchain/issues/14941

I_kwDOIPDwls56LMq8

Support tool for AzureChatOpenAI

open

NONE

### Feature request There is a new implementation of function call which I think isn't supported by langchain yet. https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/function-calling ### Motivation AzureChatOpenAI models can't be used by OpenAIFunctionAgent due to the implementation issue. ### Your contribution I've implemented a workaround here. Hoping for a full solution. ```python from langchain.chat_models import AzureChatOpenAI class AzureChatOpenAIWithTooling(AzureChatOpenAI): """AzureChatOpenAI with a patch to support functions. Function calling: https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/function-calling Currently only a single function call is supported. If multiple function calls are returned by the model, only the first one is used. """ def _generate(self, messages, stop=None, run_manager=None, stream=None, **kwargs): if "functions" in kwargs: kwargs["tools"] = [ {"type": "function", "function": f} for f in kwargs.pop("functions") ] return super()._generate(messages, stop, run_manager, stream, **kwargs) def _create_message_dicts(self, messages, stop): dicts, params = super()._create_message_dicts(messages, stop) latest_call_id = {} for d in dicts: if "function_call" in d: # Record the ID for future use latest_call_id[d["function_call"]["name"]] = d["function_call"]["id"] # Convert back to tool call d["tool_calls"] = [ { "id": d["function_call"]["id"], "function": { k: v for k, v in d["function_call"].items() if k != "id" }, "type": "function", } ] d.pop("function_call") if d["role"] == "function": # Renaming as tool d["role"] = "tool" d["tool_call_id"] = latest_call_id[d["name"]] return dicts, params def _create_chat_result(self, response): result = super()._create_chat_result(response) for generation in result.generations: if generation.message.additional_kwargs.get("tool_calls"): function_calls = [ {**t["function"], "id": t["id"]} for t in generation.message.additional_kwargs.pop("tool_calls") ] # Only consider the first one. generation.message.additional_kwargs["function_call"] = function_calls[ 0 ] return result ``` Test code: ```python def test_azure_chat_openai(): from scripts.aoai_llm import AzureChatOpenAIWithTooling agent = OpenAIFunctionsAgent.from_llm_and_tools( llm=AzureChatOpenAIWithTooling(azure_deployment="gpt-35-turbo", api_version="2023-12-01-preview", temperature=0.), tools=[ StructuredTool.from_function(get_current_weather) ], ) action = agent.plan([], input="What's the weather like in San Francisco?") print(action) tool_output = get_current_weather(**action.tool_input) result = agent.plan([ (action, tool_output) ], input="What's the weather like in San Francisco?") print(result) # Example function hard coded to return the same weather # In production, this could be your backend API or an external API def get_current_weather(location: str, unit: str = "fahrenheit"): """Get the current weather in a given location""" if "tokyo" in location.lower(): return json.dumps({"location": "Tokyo", "temperature": "10", "unit": unit}) elif "san francisco" in location.lower(): return json.dumps( {"location": "San Francisco", "temperature": "72", "unit": unit} ) elif "paris" in location.lower(): return json.dumps({"location": "Paris", "temperature": "22", "unit": unit}) else: return json.dumps({"location": location, "temperature": "unknown"}) ``` (Note: the original example to ask about weather in three countries simultaneously doesn't work here.)

https://api.github.com/repos/langchain-ai/langchain/issues/14941/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14940

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14940/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14940/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14940/events

https://github.com/langchain-ai/langchain/pull/14940

PR_kwDOIPDwls5ibac3

docs: links

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14940/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14939

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14939/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14939/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14939/events

https://github.com/langchain-ai/langchain/issues/14939

I_kwDOIPDwls56LJOL

Embeddings

open

NONE

### System Info Traceback (most recent call last): File "c:\Users\vivek\OneDrive\Desktop\Hackathon\doc.py", line 43, in <module> db = FAISS.from_documents(documents=pages, embedding=embeddings) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\schema\vectorstore.py", line 510, in from_documents return cls.from_texts(texts, embedding, metadatas=metadatas, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\vectorstores\faiss.py", line 911, in from_texts embeddings = embedding.embed_documents(texts) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\embeddings\openai.py", line 549, in embed_documents return self._get_len_safe_embeddings(texts, engine=engine) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\embeddings\openai.py", line 392, in _get_len_safe_embeddings encoding = tiktoken.encoding_for_model(model_name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\tiktoken\model.py", line 75, in encoding_for_model return get_encoding(encoding_name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\tiktoken\registry.py", line 63, in get_encoding enc = Encoding(**constructor()) ^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\tiktoken_ext\openai_public.py", line 64, in cl100k_base mergeable_ranks = load_tiktoken_bpe( ^^^^^^^^^^^^^^^^^^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\tiktoken\load.py", line 115, in load_tiktoken_bpe return { ^ File "C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\tiktoken\load.py", line 117, in <dictcomp> for token, rank in (line.split() for line in contents.splitlines() if line) ^^^^^^^^^^^ ValueError: not enough values to unpack (expected 2, got 1) ### Who can help? _No response_ ### Information - [ ] The official example notebooks/scripts - [ ] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [X] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [X] Vector Stores / Retrievers - [X] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [X] Callbacks/Tracing - [ ] Async ### Reproduction from langchain.document_loaders import PyPDFLoader from langchain.embeddings.openai import OpenAIEmbeddings from langchain.embeddings import AzureOpenAIEmbeddings from langchain.vectorstores import FAISS from dotenv import load_dotenv import openai import os #load environment variables load_dotenv() OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") OPENAI_DEPLOYMENT_ENDPOINT = os.getenv("OPENAI_DEPLOYMENT_ENDPOINT") OPENAI_DEPLOYMENT_NAME = os.getenv("OPENAI_DEPLOYMENT_NAME") OPENAI_MODEL_NAME = os.getenv("OPENAI_MODEL_NAME") OPENAI_DEPLOYMENT_VERSION = os.getenv("OPENAI_DEPLOYMENT_VERSION") OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME = os.getenv("OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME") OPENAI_ADA_EMBEDDING_MODEL_NAME = os.getenv("OPENAI_ADA_EMBEDDING_MODEL_NAME") #init Azure OpenAI openai.api_type = "azure" openai.api_version = OPENAI_DEPLOYMENT_VERSION openai.api_base = OPENAI_DEPLOYMENT_ENDPOINT openai.api_key = OPENAI_API_KEY # if __name__ == "__main__": embeddings=AzureOpenAIEmbeddings(deployment=OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME, model=OPENAI_ADA_EMBEDDING_MODEL_NAME, azure_endpoint=OPENAI_DEPLOYMENT_ENDPOINT, openai_api_type="azure", chunk_size=100) # dataPath = "./data/documentation/" fileName = r'C:\Users\vivek\OneDrive\Desktop\Hackathon\data\FAQ For LTO Hotels.pdf' #use langchain PDF loader loader = PyPDFLoader(fileName) #split the document into chunks pages = loader.load_and_split() #Use Langchain to create the embeddings using text-embedding-ada-002 db = FAISS.from_documents(documents=pages, embedding=embeddings) #save the embeddings into FAISS vector store db.save_local(r"C:\Users\vivek\OneDrive\Desktop\Hackathon\index") from dotenv import load_dotenv import os import openai from langchain.chains import RetrievalQA from langchain.vectorstores import FAISS from langchain.chains.question_answering import load_qa_chain from langchain.chat_models import AzureChatOpenAI from langchain.embeddings.openai import OpenAIEmbeddings from langchain.vectorstores import FAISS from langchain.chains import ConversationalRetrievalChain from langchain.prompts import PromptTemplate #load environment variables load_dotenv() OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") OPENAI_DEPLOYMENT_ENDPOINT = os.getenv("OPENAI_DEPLOYMENT_ENDPOINT") OPENAI_DEPLOYMENT_NAME = os.getenv("OPENAI_DEPLOYMENT_NAME") OPENAI_MODEL_NAME = os.getenv("OPENAI_MODEL_NAME") OPENAI_DEPLOYMENT_VERSION = os.getenv("OPENAI_DEPLOYMENT_VERSION") OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME = os.getenv("OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME") OPENAI_ADA_EMBEDDING_MODEL_NAME = os.getenv("OPENAI_ADA_EMBEDDING_MODEL_NAME") def ask_question(qa, question): result = qa({"query": question}) print("Question:", question) print("Answer:", result["result"]) def ask_question_with_context(qa, question, chat_history): query = "what is Azure OpenAI Service?" result = qa({"question": question, "chat_history": chat_history}) print("answer:", result["answer"]) chat_history = [(query, result["answer"])] return chat_history if __name__ == "__main__": # Configure OpenAI API openai.api_type = "azure" openai.api_base = os.getenv('OPENAI_API_BASE') openai.api_key = os.getenv("OPENAI_API_KEY") openai.api_version = os.getenv('OPENAI_API_VERSION') llm = AzureChatOpenAI(deployment_name=OPENAI_DEPLOYMENT_NAME, model_name=OPENAI_MODEL_NAME, openai_api_base=OPENAI_DEPLOYMENT_ENDPOINT, openai_api_version=OPENAI_DEPLOYMENT_VERSION, openai_api_key=OPENAI_API_KEY, openai_api_type="azure") embeddings=AzureOpenAIEmbeddings(deployment=OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME, model=OPENAI_ADA_EMBEDDING_MODEL_NAME, azure_endpoint=OPENAI_DEPLOYMENT_ENDPOINT, openai_api_type="azure", chunk_size=100) # Initialize gpt-35-turbo and our embedding model #load the faiss vector store we saved into memory vectorStore = FAISS.load_local(r"C:\Users\vivek\OneDrive\Desktop\Hackathon\index", embeddings) #use the faiss vector store we saved to search the local document retriever = vectorStore.as_retriever(search_type="similarity", search_kwargs={"k":2}) QUESTION_PROMPT = PromptTemplate.from_template("""Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question. Chat History: {chat_history} Follow Up Input: {question} Standalone question:""") qa = ConversationalRetrievalChain.from_llm(llm=llm, retriever=retriever, condense_question_prompt=QUESTION_PROMPT, return_source_documents=True, verbose=False) chat_history = [] while True: query = input('you: ') if query == 'q': break chat_history = ask_question_with_context(qa, query, chat_history) ### Expected behavior QA

https://api.github.com/repos/langchain-ai/langchain/issues/14939/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14938

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14938/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14938/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14938/events

https://github.com/langchain-ai/langchain/pull/14938

PR_kwDOIPDwls5ibWYU

docs: remove unused contributor steps

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14938/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14937

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14937/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14937/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14937/events

https://github.com/langchain-ai/langchain/issues/14937

I_kwDOIPDwls56LEjp

agent意图识别

open

NONE

假设我基于langchain分别实现了用于数据库查询的mysqlagent、用于访问外部链接apichain、以及用于rag的知识查询agent，我如何通过用户输入，将用户的请求分发到不同的agent

https://api.github.com/repos/langchain-ai/langchain/issues/14937/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14936

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14936/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14936/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14936/events

https://github.com/langchain-ai/langchain/pull/14936

PR_kwDOIPDwls5ibN75

together: package and embedding model

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14936/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14935

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14935/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14935/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14935/events

https://github.com/langchain-ai/langchain/pull/14935

PR_kwDOIPDwls5ibK6C

Add support for Vertex ai multimodal embeddings

open

NONE

<!-- Thank you for contributing to LangChain! Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes (if applicable), - **Dependencies:** any dependencies required for this change, - **Tag maintainer:** for a quicker response, tag the relevant maintainer (see below), - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/extras` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. -->

https://api.github.com/repos/langchain-ai/langchain/issues/14935/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14934

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14934/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14934/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14934/events

https://github.com/langchain-ai/langchain/issues/14934

I_kwDOIPDwls56K45f

AZure Openai Embeddings

open

NONE

### System Info C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\embeddings\azure_openai.py:101: UserWarning: As of openai>=1.0.0, Azure endpoints should be specified via the `azure_endpoint` param not `openai_api_base` (or alias `base_url`). Updating `openai_api_base` from <your openai endpoint> to <your openai endpoint>/openai. warnings.warn( C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\embeddings\azure_openai.py:108: UserWarning: As of openai>=1.0.0, if `deployment` (or alias `azure_deployment`) is specified then `openai_api_base` (or alias `base_url`) should not be. Instead use `deployment` (or alias `azure_deployment`) and `azure_endpoint`. warnings.warn( C:\Users\vivek\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\embeddings\azure_openai.py:116: UserWarning: As of openai>=1.0.0, if `openai_api_base` (or alias `base_url`) is specified it is expected to be of the form https://example-resource.azure.openai.com/openai/deployments/example-deployment. Updating <your openai endpoint> to <your openai endpoint>/openai. warnings.warn( Traceback (most recent call last): File "c:\Users\vivek\OneDrive\Desktop\Hackathon\doc.py", line 28, in <module> embeddings=AzureOpenAIEmbeddings(deployment=OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "pydantic\main.py", line 341, in pydantic.main.BaseModel.__init__ pydantic.error_wrappers.ValidationError: 1 validation error for AzureOpenAIEmbeddings __root__ base_url and azure_endpoint are mutually exclusive (type=value_error) ### Who can help? _No response_ ### Information - [ ] The official example notebooks/scripts - [ ] My own modified scripts ### Related Components - [X] LLMs/Chat Models - [X] Embedding Models - [ ] Prompts / Prompt Templates / Prompt Selectors - [ ] Output Parsers - [ ] Document Loaders - [X] Vector Stores / Retrievers - [ ] Memory - [ ] Agents / Agent Executors - [ ] Tools / Toolkits - [ ] Chains - [ ] Callbacks/Tracing - [ ] Async ### Reproduction from langchain.document_loaders import PyPDFLoader from langchain.embeddings.openai import OpenAIEmbeddings from langchain.embeddings import AzureOpenAIEmbeddings from langchain.vectorstores import FAISS from dotenv import load_dotenv import openai import os #load environment variables load_dotenv() OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") OPENAI_DEPLOYMENT_ENDPOINT = os.getenv("OPENAI_DEPLOYMENT_ENDPOINT") OPENAI_DEPLOYMENT_NAME = os.getenv("OPENAI_DEPLOYMENT_NAME") OPENAI_MODEL_NAME = os.getenv("OPENAI_MODEL_NAME") OPENAI_DEPLOYMENT_VERSION = os.getenv("OPENAI_DEPLOYMENT_VERSION") OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME = os.getenv("OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME") OPENAI_ADA_EMBEDDING_MODEL_NAME = os.getenv("OPENAI_ADA_EMBEDDING_MODEL_NAME") #init Azure OpenAI openai.api_type = "azure" openai.api_version = OPENAI_DEPLOYMENT_VERSION openai.api_base = OPENAI_DEPLOYMENT_ENDPOINT openai.api_key = OPENAI_API_KEY # if __name__ == "__main__": embeddings=AzureOpenAIEmbeddings(deployment=OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME, model=OPENAI_ADA_EMBEDDING_MODEL_NAME, openai_api_base=OPENAI_DEPLOYMENT_ENDPOINT, openai_api_type="azure", chunk_size=100) # dataPath = "./data/documentation/" fileName = r'C:\Users\vivek\OneDrive\Desktop\Hackathon\data\FAQ For LTO Hotels.pdf' #use langchain PDF loader loader = PyPDFLoader(fileName) #split the document into chunks pages = loader.load_and_split() #Use Langchain to create the embeddings using text-embedding-ada-002 db = FAISS.from_documents(documents=pages, embedding=embeddings) #save the embeddings into FAISS vector store db.save_local(r"C:\Users\vivek\OneDrive\Desktop\Hackathon\index") from dotenv import load_dotenv import os import openai from langchain.chains import RetrievalQA from langchain.vectorstores import FAISS from langchain.chains.question_answering import load_qa_chain from langchain.chat_models import AzureChatOpenAI from langchain.embeddings.openai import OpenAIEmbeddings from langchain.vectorstores import FAISS from langchain.chains import ConversationalRetrievalChain from langchain.prompts import PromptTemplate #load environment variables load_dotenv() OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") OPENAI_DEPLOYMENT_ENDPOINT = os.getenv("OPENAI_DEPLOYMENT_ENDPOINT") OPENAI_DEPLOYMENT_NAME = os.getenv("OPENAI_DEPLOYMENT_NAME") OPENAI_MODEL_NAME = os.getenv("OPENAI_MODEL_NAME") OPENAI_DEPLOYMENT_VERSION = os.getenv("OPENAI_DEPLOYMENT_VERSION") OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME = os.getenv("OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME") OPENAI_ADA_EMBEDDING_MODEL_NAME = os.getenv("OPENAI_ADA_EMBEDDING_MODEL_NAME") def ask_question(qa, question): result = qa({"query": question}) print("Question:", question) print("Answer:", result["result"]) def ask_question_with_context(qa, question, chat_history): query = "what is Azure OpenAI Service?" result = qa({"question": question, "chat_history": chat_history}) print("answer:", result["answer"]) chat_history = [(query, result["answer"])] return chat_history if __name__ == "__main__": # Configure OpenAI API openai.api_type = "azure" openai.api_base = os.getenv('OPENAI_API_BASE') openai.api_key = os.getenv("OPENAI_API_KEY") openai.api_version = os.getenv('OPENAI_API_VERSION') llm = AzureChatOpenAI(deployment_name=OPENAI_DEPLOYMENT_NAME, model_name=OPENAI_MODEL_NAME, openai_api_base=OPENAI_DEPLOYMENT_ENDPOINT, openai_api_version=OPENAI_DEPLOYMENT_VERSION, openai_api_key=OPENAI_API_KEY, openai_api_type="azure") embeddings=OpenAIEmbeddings(deployment=OPENAI_ADA_EMBEDDING_DEPLOYMENT_NAME, model=OPENAI_ADA_EMBEDDING_MODEL_NAME, openai_api_base=OPENAI_DEPLOYMENT_ENDPOINT, openai_api_type="azure", chunk_size=1) # Initialize gpt-35-turbo and our embedding model #load the faiss vector store we saved into memory vectorStore = FAISS.load_local(r"C:\Users\vivek\OneDrive\Desktop\Hackathon\index", embeddings) #use the faiss vector store we saved to search the local document retriever = vectorStore.as_retriever(search_type="similarity", search_kwargs={"k":2}) QUESTION_PROMPT = PromptTemplate.from_template("""Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question. Chat History: {chat_history} Follow Up Input: {question} Standalone question:""") qa = ConversationalRetrievalChain.from_llm(llm=llm, retriever=retriever, condense_question_prompt=QUESTION_PROMPT, return_source_documents=True, verbose=False) chat_history = [] while True: query = input('you: ') if query == 'q': break chat_history = ask_question_with_context(qa, query, chat_history) ### Expected behavior QA

https://api.github.com/repos/langchain-ai/langchain/issues/14934/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14933

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14933/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14933/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14933/events

https://github.com/langchain-ai/langchain/pull/14933

PR_kwDOIPDwls5ibJv8

community[patch]: Add param "task" to Databricks LLM to work around serialization of transform_output_fn

closed

CONTRIBUTOR

**What is the reproduce code?** ```python from langchain.chains import LLMChain, load_chain from langchain.llms import Databricks from langchain.prompts import PromptTemplate def transform_output(response): # Extract the answer from the responses. return str(response["candidates"][0]["text"]) def transform_input(**request): full_prompt = f"""{request["prompt"]} Be Concise. """ request["prompt"] = full_prompt return request chat_model = Databricks( endpoint_name="llama2-13B-chat-Brambles", transform_input_fn=transform_input, transform_output_fn=transform_output, verbose=True, ) print(f"Test chat model: {chat_model('What is Apache Spark')}") # This works llm_chain = LLMChain(llm=chat_model, prompt=PromptTemplate.from_template("{chat_input}")) llm_chain("colorful socks") # this works llm_chain.save("databricks_llm_chain.yaml") # transform_input_fn and transform_output_fn are not serialized into the model yaml file loaded_chain = load_chain("databricks_llm_chain.yaml") # The Databricks LLM is recreated with transform_input_fn=None, transform_output_fn=None. loaded_chain("colorful socks") # Thus this errors. The transform_output_fn is needed to produce the correct output ``` Error: ``` File "/local_disk0/.ephemeral_nfs/envs/pythonEnv-6c34afab-3473-421d-877f-1ef18930ef4d/lib/python3.10/site-packages/pydantic/v1/main.py", line 341, in __init__ raise validation_error pydantic.v1.error_wrappers.ValidationError: 1 validation error for Generation text str type expected (type=type_error.str) request payload: {'query': 'What is a databricks notebook?'}'} ``` **What does the error mean?** When the LLM generates an answer, represented by a Generation data object. The Generation data object takes a str field called text, e.g. Generation(text=”blah”). However, the Databricks LLM tried to put a non-str to text, e.g. Generation(text={“candidates”:[{“text”: “blah”}]}) Thus, pydantic errors. **Why the output format becomes incorrect after saving and loading the Databricks LLM?** Databrick LLM does not support serializing transform_input_fn and transform_output_fn, so they are not serialized into the model yaml file. When the Databricks LLM is loaded, it is recreated with transform_input_fn=None, transform_output_fn=None. Without transform_output_fn, the output text is not unwrapped, thus errors. Missing transform_output_fn causes this error. Missing transform_input_fn causes the additional prompt “Be Concise.” to be lost after saving and loading. <!-- Thank you for contributing to LangChain! Replace this entire comment with: - **Description:** a description of the change, - **Issue:** the issue # it fixes (if applicable), - **Dependencies:** any dependencies required for this change, - **Tag maintainer:** for a quicker response, tag the relevant maintainer (see below), - **Twitter handle:** we announce bigger features on Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out! Please make sure your PR is passing linting and testing before submitting. Run `make format`, `make lint` and `make test` to check this locally. See contribution guidelines for more information on how to write/run tests, lint, etc: https://python.langchain.com/docs/contributing/ If you're adding a new integration, please include: 1. a test for the integration, preferably unit tests that do not rely on network access, 2. an example notebook showing its use. It lives in `docs/extras` directory. If no one reviews your PR within a few days, please @-mention one of @baskaryan, @eyurtsev, @hwchase17. -->

https://api.github.com/repos/langchain-ai/langchain/issues/14933/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14931

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14931/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14931/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14931/events

https://github.com/langchain-ai/langchain/pull/14931

PR_kwDOIPDwls5iaqNU

Add langsmith and benchmark repo links

closed

COLLABORATOR

Think we could link to these in more places

https://api.github.com/repos/langchain-ai/langchain/issues/14931/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14930

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14930/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14930/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14930/events

https://github.com/langchain-ai/langchain/pull/14930

PR_kwDOIPDwls5iai9j

community[patch]: Matching engine, return doc id

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14930/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14929

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14929/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14929/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14929/events

https://github.com/langchain-ai/langchain/pull/14929

PR_kwDOIPDwls5iaePv

Update embedding_distance.ipynb

closed

CONTRIBUTOR

**Description:** Fix the docs about embedding distance evaluations guide.

https://api.github.com/repos/langchain-ai/langchain/issues/14929/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14928

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14928/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14928/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14928/events

https://github.com/langchain-ai/langchain/pull/14928

PR_kwDOIPDwls5iacaa

anthropic: beta messages integration

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14928/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14927

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14927/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14927/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14927/events

https://github.com/langchain-ai/langchain/pull/14927

PR_kwDOIPDwls5iaZrK

community[minor]: Add HF chat models

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14927/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14926

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14926/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14926/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14926/events

https://github.com/langchain-ai/langchain/pull/14926

PR_kwDOIPDwls5iaVU-

ConvoOutputParser: Handle multiline Action Input + potential hallucin…

closed

CONTRIBUTOR

…ated Observation - **Description:** Handle multiline Action Inputs. - **Issue:** Fixes cases where our agent made multiline Action Input (specifically contained python code) - **Dependencies:** -- - **Tag maintainer:** @hwchase17 - **Twitter handle:** All the credits to @robocorp Note: I did not fully get where the tests should go so that is my best effort.

https://api.github.com/repos/langchain-ai/langchain/issues/14926/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14925

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14925/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14925/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14925/events

https://github.com/langchain-ai/langchain/pull/14925

PR_kwDOIPDwls5iaHcY

docs `huggingface` platform page update

closed

COLLABORATOR

Added a reference to the `Hugging Face prompt injection identification` page

https://api.github.com/repos/langchain-ai/langchain/issues/14925/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14924

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14924/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14924/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14924/events

https://github.com/langchain-ai/langchain/pull/14924

PR_kwDOIPDwls5iaE6z

cli: test_integration group

closed

COLLABORATOR

https://api.github.com/repos/langchain-ai/langchain/issues/14924/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14923

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14923/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14923/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14923/events

https://github.com/langchain-ai/langchain/pull/14923

PR_kwDOIPDwls5iaBoC

Don't reassign chunk_type

closed

CONTRIBUTOR

**Description**: The parameter chunk_type was being hard coded to "extractive_answers", so that when "snippet" was being passed, it was being ignored. This change simply doesn't do that.

https://api.github.com/repos/langchain-ai/langchain/issues/14923/timeline

https://api.github.com/repos/langchain-ai/langchain/issues/14922

https://api.github.com/repos/langchain-ai/langchain

https://api.github.com/repos/langchain-ai/langchain/issues/14922/labels{/name}

https://api.github.com/repos/langchain-ai/langchain/issues/14922/comments

https://api.github.com/repos/langchain-ai/langchain/issues/14922/events

https://github.com/langchain-ai/langchain/pull/14922

PR_kwDOIPDwls5iZxDJ

Update google_cloud_storage_directory.ipynb

closed

CONTRIBUTOR

- Description: Just a minor add to the documentation to clarify how to load all files from a folder. I assumed and try to do it specifying it in the bucket (BUCKET/FOLDER), instead of using the prefix.

https://api.github.com/repos/langchain-ai/langchain/issues/14922/timeline