lm_understand-logs

Currently trending because

The authors report that GPT-6 Astra edited 0.016% of Qwen3.5-0.8B’s weights—no gradients—and MATH-500 rose from 7.9% to 48.1%, largely by stopping a reasoning loop that ate the token budget.^[1]^[3]

Daily observations from the public event archive.

Daily stars & forks

Loading attention chart

Cumulative stars

Loading star accumulation chart

Daily opens & merges

Loading pull requests chart

Daily issues & comments

Loading conversations chart

Daily accounts & reviews

Loading community chart

Daily releases

Loading releases chart

The last 10 distinct accounts we observed this year.

PR, issue, comment & review events in the past year.

No packages found.

Embed a badge in your README or website.

Live mentions · last 30 days