Huginn-style looped models beat Ouro at matched scale
A Huginn-style MoE uses 75% fewer resident parameters than a 112-layer feedforward baseline under matched FLOPs.
42
- Mentions · 7d0
- Mentions · 30d5
- All mentions5
- First seenSep 30, 2026
- Last mentioned10d ago
Timeline
5 articles · newest firstTowards Looped Models Done Right, Part II: Rethinking at Fixed Points
Looped language models near fixed points cut training and decode cost via learned depth priors and orthogonal injection.
44
Towards Looped Models Done Right, Part II: Rethinking at Fixed Points
Learned depth priors and orthogonal injection improve looped language models while enabling cheaper KV sharing.
50
Decoding Looped Transformers Better for (Almost) Free
Training-free LoopCD improves looped Transformer decoding and can cut recurrent compute by up to 48%.
48
Decoding Looped Transformers Better for (Almost) Free
LoopCD uses earlier recurrent states for contrastive decoding, boosting looped Transformers while cutting inference FLOPs.
55
Appears with
- Ouro-2.6B-Thinkingai model · 2
- Ouroai model · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.