CWM brings white-box adaptive RAG to SOTA on three benchmarks at EMNLP 2026
_reachsumit · x · 2026-09-15
CWM (Controllable White-Box Meta-Prompting) uses internal input-level confidence signals to decide when to retrieve, replacing black-box adaptive RAG methods without external decision modules or multi-sampling. It hits SOTA on three adaptive RAG benchmarks across GPT-oss-20b, Qwen3-14b, and Llama3.1-8b, generalizes to reasoning tasks, and offers controllability via internal signal manipulation. Code is open-sourced.
More from Research
- BrickWright: agents build LEGO assemblies with perfect physical validity — cmu-gil · 2026-10-24
- A byte-level transformer works, but BPE exists for a reason: context length — max_paperclips · 2026-10-24
- Community note debunks claim that distant tokens never share MoE experts — NathanpmYoung · 2026-10-24
- Many LLM users never touched classic classifiers — old-school ML hygiene is fading — antoine_chaffin · 2026-10-24
- Google shows neural cellular automata can do multi-step reasoning on decentralized hardware — tyrell_turing · 2026-10-24
- Revisiting Dario's 'Big Blob of Compute': RL distributions teach no learnable alignment lesson — 1a3orn · 2026-10-24