feat(data): resolve multitaste via the Taste column; keep sour/salty as rules - #33
Merged
Merged
Conversation
…as rules Parse ChemTastesDB's granular 'Taste' column (Sweet/Bitter, Non-sweet; Bitter, Salty/Umami, Slightly bitter, ...) and merge it positive-wins with the coarse 'Class taste'. Resolves 172/176 multitaste rows into concrete labels and adds ~400 positives (sweet 1204->1371, bitter 1563->1744, umami 220->283). The extra data pushed sour over the train threshold, but sour/salty stay validated RULES by design (solution/pH and cation properties — what the API contract specifies), via an explicit RULE_TASTES guard, not a data threshold. Also clear stale model/onnx artifacts each run so a now-ruled taste can't leave an orphan head (export was picking up a ghost sour.onnx). Verified: sweet/bitter/umami AUROC 0.95/0.95/0.99; ONNX export clean (3 heads, self-validates <1e-6).
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Parse ChemTastesDB's granular
Tastecolumn and merge positive-wins withClass taste. Resolves 172/176 multitaste rows into concrete labels (+~400 positives: sweet 1204→1371, bitter 1563→1744, umami 220→283).The extra data pushed sour over the train threshold, so I added an explicit
RULE_TASTESguard — sour/salty stay rules by design (solution/pH + cation properties, per the API contract), not data-volume-gated. Also clears stale model/onnx artifacts each run (export was picking up a ghostsour.onnx).Verified end to end: sweet/bitter/umami AUROC 0.95/0.95/0.99; clean ONNX export (3 heads, <1e-6).