cmc-csci040
cmc-csci040 copied to clipboard
mikeizbicki
Project Questions
I have a few questions regarding the project and would appreciate some help:
- When we use API to score chunk for us, how should we treat our four required doctests? The output will vary.
- For ' def find_relevant_chunks(text, query, num_chunks=5)' could you provide me with a few doctest examples? I'm having trouble writing appropriate doctests, because I don't have a function that split sentences, so the text I provide actually returns the same output. Anyway to circumvent that issue?
- Do we need doctests for the extra points (TTS and STT)?
Thanks!
Sean
When we use API to score chunk for us, how should we treat our four required doctests? The output will vary.
My solution to the score_chunk function from class did not use the LLM API and therefore was not nondeterministic.
For ' def find_relevant_chunks(text, query, num_chunks=5)' could you provide me with a few doctest examples? I'm having trouble writing appropriate doctests, because I don't have a function that split sentences, so the text I provide actually returns the same output. Anyway to circumvent that issue?
I'm sorry I don't understand the question. The chunk_text_by_words function that you're required to write splits the the text into chunks.
Do we need doctests for the extra points (TTS and STT)?
You may create one function per extra credit that does not have doctests. Any other functions created need doctests.
Hey Mike,
Sure, I see. For the extra points, how should we write the doctests though? TTS does not have a written output, and STT does not have a written input. I consulted with ChatGPT and it says I might consider using '_' to ignore the output, and just put the doctests as references. Do you think that works? Or do you have specific expectations?
You can write one function for TTS and one for STT that doesn't have doctests. If you want to write more than these 2 functions, then the other functions will need doctests.