Loom lens: sampling the MTP head like a language model to get tree-shaped activation readouts
voooooogel · x · 2026-11-12
vgel experiments with the "loom lens": since an MTP head can be sampled like a regular language model, hidden-state readouts can be rendered as branching trees of multi-token continuations. It's not yet better than plain chain lens — the tree structure confuses the judge — but the team thinks it has potential in the hands of an interpretation agent.
Related event: vgel team explores Loom Lens: sampling MTP heads like a language model(2 posts)→
More from Research
- A LoRA Is All You Need to Turn a Speech LLM Full-Duplex — kastnerkyle · 2026-11-12
- WICL: robots learn manipulation in-context, 83.4% success on all 50 RoboTwin tasks — _akhaliq · 2026-11-12
- Epoch: AI software progress cuts training compute needs several-fold per year — datagenproc · 2026-11-12
- Blood on the Clocktower benchmark: Kimi K3.1 Pro tops open models, just below GPT 5.5 — eliebakouch · 2026-11-12
- Gemini runs Blood on the Clock Tower benchmark nearly autonomously in new work — eliebakouch · 2026-11-12
- Replicating all 9 weight-space programming experiments: big wins, one miss — burny_tech · 2026-11-12