GitHub×HF agent RL workshop oversubscribed 5x; host refuses to expand capacity
ben_burtenshaw · x · 2026-10-14
The "Training Agents IRL" workshop hosted by Ben Burtenshaw at GitHub's SF office is 5x oversubscribed, but he won't increase capacity, preferring a focused community event and asking no-shows to drop out.
Co-hosted by GitHub and Hugging Face as part of Open Source AI Week, it focuses on RL for agentic systems: reward design, rollouts, benchmarks, and the gap between training gains and real-world agent behavior, with speakers from HF, Microsoft, Mercor, and Reflection in 10-minute talks plus a closing panel.
More from Companies & People
- UNC CS opens tenure-track hires in AI Safety and Neuroengineering, review starts Dec 1 — mohitban47 · 2026-10-14
- OpenAI Names Former xAI Exec Uday Ruddarraju to Lead Infra and Compute — dinabass · 2026-10-14
- Robot foundation-model startups are all opening APIs — the model layer was never the moat — chris_j_paxton · 2026-10-14
- Research intern hiring season: everyone wants interns who are already PhD students — k7agar · 2026-10-14
- Vector Scholarship in AI rises to CAD $21,000 with direct industry access — VectorInst · 2026-10-14
- Shreyas: top 20% of product builders already talk to users enough — judgment matters most — annetgriffin · 2026-10-14