Assignment#2
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changes are you trying to make? (e.g. Adding or removing code, refactoring existing code, adding reports)
What did you learn from the changes you have made?
How to load env, use langchain and deepeval for evaluation, and extract data from pdf.
Was there another approach you were thinking about making? If so, what approach(es) were you thinking of?
Yeah, a few times I went back and forth before settling on what's in the notebook. Here's roughly how that went
the enhancement step, I thought about running two or three rounds of self-correction instead of just one basically having the model critique its own critic but I held off. Partly scope, partly because what I actually saw (the enhanced summary scoring worse than the original on a couple of metrics) made me suspicious that more rounds would just compound that problem rather than fix it. That's actually the approach I'd want to try next if I kept going: instead of one enhancement pass, average several samples at each stage and only keep a revision if it wins on a majority of metrics, rather than trusting a single before/after comparison.
Were there any challenges? If so, what issue(s) did you face? How did you overcome it?
Yes , env setup was a challenge and first assignment submission was also challenge .
How were these changes tested?
In jupyternotebook
A reference to a related issue in your repository (if applicable)
Checklist