Hi there! I took the instructor led Building RAG Agents With LLMs last week on October 7th 2024 and I am stuck on the very last part in this course. In the last part we are supposed to evaluate our RAG by setting up a langserve server and showing it on gradios. I followed the instructions from a previous notebook on how to set up the langserve server but nothing is working. I’ve tried to change the endpoints, I’ve tried to change the coding to allow it to work properly (without nest_asyncio function I won’t be able to run the server), I even used online resources like chatgpt to help fix the issue.
Everything I’ve done ended up as an error and now I’ve used up all my timing to finish the course. I’ve tried to contacted dli-help@nvidia.com multiple times about this issue but they have not contacted me for a week and just sent me an email offering me a 50% discount on their next workshop.
Please help out with this issue I’d like to better understand what the issue could be.
Here’s a list of errors I’ve encountered from this:
INFO: 127.0.0.1:40150 - “POST /retriever HTTP/1.1” 404 Not Found
I added some more time to your session. Regarding the issue, let me see…
Regarding nest_asyncio, the langserve deployment part actually recommends the %%writefile + !python approach which shouldn’t require any special notebook concurrency handling.
Curious to see what approach you took. The recommendation on the retriever front involves recreating the index in your file (from the index cache) and deploying the retriever form (i.e. .as_retriever()).
Please let me know how you were approaching it when you get the time. Feel free to post some code as well, or DM me directly if you want to send your entire submission.
I have a problem with the same course but during the part of evaluation .
I have completed the course 3 time with all the “TODO” In the exercises completely, but still I don’t have a single waited average needed for the course .
When ever I clicked the validate button, it always says I have not completed any task .
We recently had a student try to reconstruct the gradio frontend in their notebook and run it in their /lab microservice from the jupyter notebook. The autograder looks for a PASSED file to be written inside the frontend microservice - Hence, trying to replicate the environment outside of the frontend microservice results in a success message without the file being written. In other words, if the success message is provided, but not in the :8090 interface, the assessment should not pass.