Looped transformers: more looping boosts reasoning but with diminishing returns past 3 passes and fewer CoT tokens
nrehiew_ · x · 2026-11-21
nrehiew summarizes findings from the TML-d24-varloop-research model (variable looped layers):
- Finding 1: More looping raises no-CoT time horizons, with diminishing returns.
- Finding 2: Looping especially helps multi-hop "reasoning" tasks.
- Finding 3: Performance improves with more looping but plateaus after 3 passes; fewer placeholder tokens also means fewer reasoning tokens needed. Charts show mean reasoning tokens dropping from 3703 to 1661 at 5 passes, and uncertainty indicators ("wait", "hmm") falling from 9.7 to 3.6.
More from Models
- Gemini 4 Neon locked to Pro and Ultra paid subscribers, leaving Plus users out — giffmana · 2026-11-21
- Team releases TML-d24-varloop-research on Hugging Face to study looping's effect on CoT monitorability — nrehiew_ · 2026-11-21
- Full sub-benchmark breakdown for Inkling 2 Mini published by Artificial Analysis — ArtificialAnlys · 2026-11-21
- Inkling 2 Mini costs ~6x less and runs ~4x faster per task than Qwen3.8 27B, AA finds — ArtificialAnlys · 2026-11-21
- Inkling 2 Mini matches 27B-33B rivals on intelligence with ~5-6x fewer active parameters — ArtificialAnlys · 2026-11-21
- Thinking Machines' open-weights Inkling 2 Mini hits 20 on AA Intelligence Index with just 5.4B active params — ArtificialAnlys · 2026-11-21