An input-size limit changes the request count
lib/jev.ts can recursively split a request's segments and questions, then merge answers. The README's single-request description is not a guarantee for a long transcript.
Inspect the supporting source ↗A browser extension that classifies YouTube caption segments with Jev, splitting requests when the segment or token budget is exceeded, paints five categories on the seek bar, and auto-skips those painted slices over the threshold.
One Choice per caption segment among content, sponsor, intro, outro, self_promo, recap, and other. Painted categories sponsor, self_promo, intro, outro, and recap skip when probability is at or above the threshold; content and other never skip.
Works on videos nobody has labeled yet; with no captions it does nothing.
The cited README and lib/jev.ts were checked for caption classification and recursive request splitting. No extension or live API was run here; the 23-video figures belong to the author, and the demo uses saved answers.
Skip likely sponsor or intro segments on a YouTube video that has captions but no crowd labels.
Material analysed:
lib/jev.ts can recursively split a request's segments and questions, then merge answers. The README's single-request description is not a guarantee for a long transcript.
Inspect the supporting source ↗The extension has no opinion when captions are absent. Its 23-video result and gateway measurements are author-reported, and the demo replays saved answers.
Inspect the supporting source ↗Useful when early coverage matters more than perfectly labeled timestamps. Prefer crowd-confirmed segments or manual seeking when false skips would hide important material.
At the pinned revision, follow the split function and check that each child carries only its matching question IDs. Compare the README demo notes with the measurement claims before reproducing them.