I am currently working on the lab “Exploring LLM Capabilities” (C1M4_Ungraded_Lab_1) in the Retrieval Augmented Generation course.
When I run the notebook cells without modifying anything, I receive an error when the notebook attempts to call the LLM.
The error occurs when the functions that call the model are executed (for example when running cells that use generate_with_single_input()).
The traceback shows the following error:
Error while calling LLM:
Unable to access non-serverless model meta-llama/Llama-3.2-3B-Instruct-Turbo.
“type”: “invalid_request_error”
“code”: “model_not_available”
From the traceback, the error occurs inside utils.py when the notebook sends a request to the course proxy endpoint.
It seems that the model
meta-llama/Llama-3.2-3B-Instruct-Turbo
may no longer be accessible through the proxy used by the lab.
I have not modified the notebook and the error appears when running the cells as provided.
Has the model endpoint changed, or is there currently an issue with the proxy used in the lab?
I have attached screenshots of the error for reference.
together.ai has deprecated a lot of models in the past days resulting in concurrent issues with labs of RAG courses.
Lab assignment 3 to 5 and all ungraded labs are updated now.
You need to click on , Reboot and Get the latest version to work on updated notebook.
I also would recommend to first delete the old file, and then reboot as to any cache issue is not encountered when running codes again.
Courses, labs are created with multiple channels, so labs gets affected when any changes.made at one end and can take multiple attempts to resolve issue.
Click l, click First reboot, then click get the latest version.
let me know if you still encounter the same error.
Also remember, the import libraries should be run down only once, if you run it twice, you will encounter again the same error(as even I experienced this).
Then I clicked clear out kernel output and re-run cells individually.
please remember that import code cells take time to run down, during which time you will notice *, but do not click or run down that cell again and again. it probably will take 30-50s
Unfortunately, still couldn’t make it work. Even after trying all the steps you described in a different browser with clean cache. It looks like restored notebook version comes from 6 days ago (according to its timestamp) and it is still referencing the model “meta-llama/Llama-3.2-3B-Instruct-Turbo”.
P.S. I’m experiencing the problems with “Exploring LLM Capabilities” (C1M4_Ungraded_Lab_1) (the same lab as the topic starter)
then looks like it is recurring new exception error similar lab 5 assignment which one of the learner experienced and when I checked even I experienced.
Unfortunately this part of issue only staff can resolve, so we will have to wait for @lucas.coutinho to look upon this tomorrow when weekday starts in his timeline.
, Reboot and Get the latest version to work on updated notebook.
