Two New Surveys Land: Continuous Diffusion LMs and Mechanistic Interpretability for AI Safety
burny_tech · x · 2026-11-11
alphaXiv highlights two new human-written surveys, arguing human curation matters more than ever amid paper noise.
- A Survey on Continuous Diffusion and Flow Language Models (Sophia Tang et al., UPenn/KAIST/UvA): organizes the field of continuous diffusion and flow language models — text generation via noise-to-continuous-representation transforms, parallel refinement with preserved dependencies, gradient-steerable generation. Covers few-step generation (flow maps, distillation), scaling to billions of parameters, connections to image/video generation, and open questions in representation learning, post-training, multimodality, and safety. Includes an Awesome list and interactive paper map.
- Understanding AI from the Inside (Fazl Barez, Tony Wu et al., with Anthropic, Oxford, Mila, Bengio, Stella Biderman): surveys mechanistic interpretability for AI safety, asking whether understanding how models work actually makes them safer, and compares mechanistic evidence with behavioral evaluation.
More from Research
- Epoch AI chart: AI cost falling ~13x a year, faster than any tech in history — sebkrier · 2026-11-11
- AI Trials Observatory launches: 93 RCTs on how AI affects mental health and beliefs — DG_Rand · 2026-11-11
- Raschka: open LLMs have abandoned full attention — hybrids and linear models now split 50/50 — rasbt · 2026-11-11
- Strogatz & Townsend's new book Big Math tackles math's Faustian bargain with AI — stevenstrogatz · 2026-11-11
- RLVR lets small models close the coding gap; Python's dynamic typing hurts pretraining — mdancho84 · 2026-11-11
- 50 researchers from ByteDance, Alibaba, Tencent release 303-page code AI field guide — mdancho84 · 2026-11-11