Interpretable Generative Models at ICML
kchonyc · x · 2026-07-15
ICML 上的可解释性模型与概念瓶颈生成模型
The post shares a现场 talk from ICML 2026, focusing on exchanges around interpretable-by-design models（可解释性内建模型）, along with a poster presented at the Compositional Learning workshop.
The core work is Single-Cell Concept Bottleneck Generative Models, which applies the "concept bottleneck" concept to single-cell generative modeling, emphasizing the combination of model interpretability and compositional learning.
More from Research
- NeurIPS 2026 workshop calls papers on on-device intelligence — YiMaTweets · 2026-07-21
- AI Security Institute says every tested model tried to cheat in cyber evaluations — connoraxiotes · 2026-07-21
- AI companies are buying old books to avoid training on AI-generated slop — CackleRooster · 2026-07-21
- Sakana says multiple diffusion models plus MCTS beat test-time scaling on coding and math — SakanaAILabs · 2026-07-21
- Soofi S 30B-A3B releases a full pretraining report and claims open-model leads in English and German — abursuc · 2026-07-21
- AI Companies Are Buying Tons of Old Books Because They're Free of AI Slop — 404 Media · 2026-07-21