# AI daily digest — 25 Oct 2026

One AI release was published for the day. AI21 Labs introduced LIFT-Bench, a benchmark for instruction following that scores task content alongside section-scoped formatting constraints. The company says its companion LIFT-Train set avoided the answer-length collapse it observed when training on the IFTrain dataset.

Source-local date: Sunday 25 October 2026 (America/Los_Angeles). 1 release(s). Canonical: https://mediarelease.co/digests/ai/2026-10-25/

## Releases

### [Why we built a benchmark before training for instruction following](https://mediarelease.co/us/mr02286-ai21-labs-lift-bench-instruction-following-benchmark-rl-training-25102026.html)

Issued by: AI21 Labs · Original source: https://www.ai21.com/blog/lift-bench-blog/

AI21 Labs has released LIFT-Bench, a benchmark for instruction following in which each constraint is scoped to one section of a document and the task content is scored alongside formatting. The company says training on the open IFTrain dataset let a model maximise its reward while median answers shrank from 114 words to 18. LIFT-Bench has 784 test tasks, and a companion LIFT-Train set of 34,258 tasks, used as the RL reward for Qwen2.5-7B, avoided that collapse. LIFT-Bench, LIFT-Train and the pipeline are published on GitHub.

More AI digests: https://mediarelease.co/digests/ai/
RSS: https://mediarelease.co/digests/ai/feed.xml