This course uses LangSmith (although it calls it LangChain Platform) to demonstrate visually what’s happening under the hood and encourages students to do the same. I tried signing up for LangSmith but it’s still in closed beta. Does anyone have an invite code I can use to get access to it?
Hi,
You don’t need LangSmith to use LangChain. To use it, just “pip install langchain” in your Python code.
Sam Witteveen has a good series of videos explaining how to use it while providing the Colab Code Notebook.
Moderator edit: You can find the video by searching on YouTube for “LangChain Basics Tutorial #1”
Yes, I understand that LangChain doesn’t need LangSmith, but LangSmith provides visual insight on all the intermediate steps that LangChain goes through. That’s what I’m hoping to get.
I agree. I’m also on the waitlist for Langsmith. It looks like an excellent low-code tool for speeding up Gen AI development.
Browsing Deeplearning.AI courses, I just found one that maybe helps.
For those waiting for LangSmith, in the meantime you can pass in verbose=True to LangChain to get detailed information of all the intermediary steps:
qa = ConversationalRetrievalChain.from_llm(
llm,
retriever=retriever,
memory=memory,
verbose=True
)
While this verbose setting is quite helpful, I’m still not able to see the LLM response in the intermediary steps. For that, I’ve set up a proxy that all the HTTP requests go through by pip installing pip_system_certs and using Fiddler’s http capturing feature to intercept requests and responses on a Windows machine. On a Mac, you should be able to do the same thing with tools like CharlesProxy.
Another cool thing you could try is using debug. It produces a lot of intermediate responses.
import langchain
langchain.debug=True
debug is even better. Thank you!
This is how I’m configuring it:
from langchain.globals import set_debug, set_verbose
set_debug(True)
set_verbose(True)