OpenSLA site goes live: one model answers what's happening, what to do, and why from sensor data
yang_yuzhe · x · 2026-10-08
The OpenSLA project page (UCLA) is live. Core thesis: today's sensor models stop at perception; OpenSLA lets a single model answer three questions from one sensor window plus individual context — what is happening, what action follows, and why. The demo shows an emergency-department scenario (30-minute record window), multi-level action prediction, zero-shot discrimination of unseen actions, generated captions with concrete numeric values recovered from signals, evidence grounding, zero-shot transfer to the external MIMIC-IV cohort, and readouts of glucose dynamics over the next two hours. Paper, code, and Hugging Face models are all available.
Related event: UCLA Team Open-Sources OpenSLA: Unifying Sensors, Language, and Action(7 posts)→
More from Embodied
- Learned latent dynamics roll forward from few observations to recover hopper state with uncertainty — hisspikeness · 2026-10-08
- Adding a pixel-reconstruction loss makes latents unable to decode hopper state — hisspikeness · 2026-10-08
- ASUS partners with Nous Research to ship Hermes Agent on RTX Spark PCs — Teknium · 2026-10-08
- RoboWorld Challenge 2026 opens: embodied world modeling across three tracks — liuziwei7 · 2026-10-08
- E2E self-driving startup Wayve lands Stellantis deal after Nissan, claims it beat Tesla to E2E — alexgkendall · 2026-10-08
- Rabbit OS3 update lets the R1 place outbound phone calls — SimonBalmain · 2026-10-08