Skip to content

Fix expected_rule in gradient and tangent_line benchmarks - #33

Merged
sp25-bai-047-wq merged 2 commits into
QuantumLogicsLabs:mainfrom
Momin0000:main
Aug 12, 2026
Merged

Fix expected_rule in gradient and tangent_line benchmarks#33
sp25-bai-047-wq merged 2 commits into
QuantumLogicsLabs:mainfrom
Momin0000:main

Conversation

@Momin0000

Copy link
Copy Markdown
Contributor

No description provided.

…rk_gradient.json and benchmark_tangent_line.json had expected_ruleset to the literal operation name ('gradient', 'tangent_line'), which doesnot exist anywhere in the 13 real RULE: tokens in vocab.json. This does notaffect today's exact-match accuracy (run_eval.py's accuracy calc is purelyexpr-comparison based, expected_rule is currently unread there), but itwould silently produce a permanent 0% the moment any rule-level accuracymetric is implemented -- unrelated to model quality.problem_generator.py's generate_gradient_diff() and generate_tangent_line_diff()already assign rule_id=7 (partial_derivative) and rule_id=0 (power_rule)respectively (see inline comments at those return statements). This alignsexpected_rule with what the model is actually trained to predict.Verified: all 50/50 rows in each file updated; cross-checked againstvocab.json's actual rule_tokens list -- zero invalid labels remain.
@vercel

vercel Bot commented Aug 12, 2026

Copy link
Copy Markdown

@Momin0000 is attempting to deploy a commit to the seno-quantum-coder's projects Team on Vercel.

A member of the Team first needs to authorize it.

@sp25-bai-047-wq
sp25-bai-047-wq merged commit 7d62d82 into QuantumLogicsLabs:main Aug 12, 2026
1 of 5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants