docs: complete Android benchmark coverage - #473
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Fill the remaining Android benchmark cells with validated timings and matching Settings proof, bringing each model to four readiness runs. Existing published runs and all iOS data stay unchanged.
Solution
Most of the diff is generated, sanitized timelines and screenshot/recording pairs. The exporter accepts hash-bound session-audit corrections from #471 after rechecking retained commands and all existing evidence requirements. Original records stay unchanged; stale hashes or unrelated validation reasons remain excluded.
Directory and worktree inventory output is omitted to avoid publishing coordinator contents or unrelated device names. Commands and their timings remain visible.
Test plan
Check the Android summary and deep links for Luna, Sol, Sonnet, and Opus; each has JavaScript/native Stim/control results and matching proof media. Existing run objects remain byte-equivalent after JSON parsing. Regression coverage exercises correction hash mismatches, command/event mismatch, wrong sessions, unrelated validation reasons, and unchanged original records.
Fixes #472.