Team open-sources variable-looping model to test looping's effect on CoT monitorability
Researchers trained and open-sourced TML-d24-varloop-research to test whether looping hurts CoT monitorability, finding looping helps no-CoT and multi-hop tasks most, with inference windows rising from 16 to 37 seconds.
2026-11-21 ~ 2026-11-21 · 3 related posts
- Team releases TML-d24-varloop-research on Hugging Face to study looping's effect on CoT monitorability — nrehiew_ · 2026-11-21
- Looped transformer lifts no-CoT time horizon from 16s to 37s, biggest gains on multi-hop reasoning — nrehiew_ · 2026-11-21
- Thinky open-sources variable-loop model: more looping passes boost long-context but not knowledge benchmarks — simonguozirui · 2026-11-21