NNsight 0.8 pre-release ships new engine for near-native vLLM interpretability
gsarti_ · x · 2026-09-10
NNsight 0.8 (pre-release) arrives with a new execution engine behind the same API. It supports every HF Transformers task, brings vLLM interpretability at near-native throughput, and adds an engine built for intervention performance and flexibility — moving white-box inference closer to SOTA inference engines. The upgrade previously enabled the Aletheia lie detection challenge, with the official NDIF release coming soon.
Related event: NNsight 0.8 Pre-release Brings Near-Native Throughput White-Box Inference(2 posts)→
More from Research
- PKU paper: rewriting influential training samples shifts model behavior more than reweighting — PKU · 2026-09-10
- StochBench: 450 graduate-level stochastic process problems in Lean 4 for theorem proving — Idan Davidovich · 2026-09-10
- KaRMA: first metric to quantify robot hand dexterity from kinematics alone, at IROS 2026 — pulkitology · 2026-09-10
- Ego-Exo 4D dataset highlighted as key to closing the 4D understanding gap — ducha_aiki · 2026-09-10
- Lean gets con-leche: an external checker proven consistent, never accepts a proof of False — AlexKontorovich · 2026-09-10
- Kristen Grauman talks machine perception to human expertise at ECCV 2026 — ducha_aiki · 2026-09-10