Add Monte Carlo sampling mode to SaliencyEngine#4
Merged
Conversation
n_samples>1 switches from logprob-drop scoring to sampling: each masked context is run n_samples times at sample_temperature and importance = P(original | full) - P(original | masked). Produces a genuine probability distribution instead of near-binary scores when the model is confident at temperature=0. - 11 new tests covering sampling mode, call counts, and edge cases - Demo updated to use n_samples=5 with a loan review scenario
Welcome to Codecov 🎉Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests. Thanks for integrating Codecov - We've got you covered ☂️ |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
n_samplesandsample_temperatureparameters toSaliencyEnginen_samples > 1, switches from logprob-drop scoring to Monte Carlo sampling: each masked context is runn_samplestimes atsample_temperatureand importance = P(original | full) − P(original | masked)n_samples=5) instead of near-binary scores when the model is highly confident at temperature=0n_samples=5, showing a real spread across credit score, employment, loan amount, and income segmentsTest plan
uv run pytest tests/)examples/demo_saliency.pyto verify distribution output