David Duvenaud: gradient-free 'poor-man's RL' on scaffolds works but is far slower
DavidDuvenaud · x · 2026-10-31
In a technical exchange about test-time compute scaling, David Duvenaud notes that even when training plans disallow RL rollouts for models being trained, one can still do a 'poor-man's RL' via gradient-free tuning of context and scaffolds — though he concedes it remains far slower than native RL.
More from Research
- After NVIDIA's BindClaw, every AI4Bio paper now needs an agentic pipeline demo to prove versatility — rishabh16_ · 2026-10-31
- Neural Thickets Revisited: Frozen-Model Logit Perturbations Match Full-Weight Perturbation Ensembles — kastnerkyle · 2026-10-31
- CS312 course post explains sharpness and the edge of stability in LLM pretraining — tatsu_hashimoto · 2026-10-31
- AI decodes wolf howls: 500K hours of Yellowstone audio analyzed by models — aza · 2026-10-31
- NBER paper: training your AI replacement can actually raise workers' pay — joshgans · 2026-10-31
- Trata's Hedge-Bench: no model tops 50% on real-world financial forecasting — ycombinator · 2026-10-31