docs/how_to/qa_sources/

[giscus[bot]]

docs/how_to/qa_sources/

Often in Q&A applications it's important to show users the sources that were used to generate the answer. The simplest way to do this is for the chain to return the Documents that were retrieved in each generation.

https://python.langchain.com/docs/how_to/qa_sources/

[drew-wks]

Would be helpful to know what version of LangChain this tutorial assumes, as many of the approaches have been deprecated over the past year.

[eyurtsev]

The most recent docs assume the most recent minor version of langchain. So this tutorial assumes that you're on version 0.3.x of langchain-core / langchain / langchain-community

[drew-wks]

thank you.

[SharoMonk]

There are too many repetitions of the concepts, they are all scattered all round the documentation. I will suggest you harmonise all the concepts discussed under a more representative sections and headings.

For individuals like me that reads the documentation all through for a better insights into the working of the library, y'all have done a very good work but can do better in the area i have indicated above. I honestly don't mind helping with this provided you get what i am saying.

Huge thanks to the team.

[Youbiquitous]

Am I missing something or the possibility of streaming the full chain output is not covered here? I see you're streaming the tokens or the entire state updates, but what about streaming the full LCEL chain output in the formatted form? I.e. returning {
"answer": "Chain of Thought (CoT) is a prompting technique that enhances model performance by instructing it to think step by step, allowing the decomposition of complex tasks into smaller, manageable steps. This method not only aids in task execution but also provides insights into the model's reasoning process. CoT has become a standard approach in improving how language models handle intricate problem-solving tasks.",
"sources": [
"Wei et al. 2022"
]
} token by token in the formatted form.

[OssamaEls]

I think it should be graph_builder = StateGraph(State) and not graph_builder = StateGraph(MessageState), otherwise I don't get the context key in the step dict.