giffmana Pushes Back: Does RL Work Without Pretraining in Tabular Settings?
giffmana · x · 2026-11-08
giffmana challenges DimitrisPapail's claim that RL doesn't work without pretraining by pointing to the tabular setting, where training from scratch is feasible. Papail had trained his own model from scratch and concluded RL fails without pretraining.
Related event: Researcher Finds RL Fails Without Pretraining, Sparking Debate(2 posts)→
More from Research
- Inkling 2 Small notes: 594B-param MoE with 17B active, RL ate ~half of pretraining FLOPs — teortaxesTex · 2026-11-08
- Looped Transformers as Programmable Computers: a 13-layer transformer emulating a full instruction-set CPU — DimitrisPapail · 2026-11-08
- New Science paper by James Evans and Glen Weyl: institutions, not models, are AI's binding constraint — round · 2026-11-08
- One BOS token changed 6 of 15 LLM responses in a controlled experiment — Full-Travel857 · 2026-11-08
- LittlePly chess agent project publishes its paper and repository — andreisavu · 2026-11-08
- LittlePly's analysis lab exposes per-move model estimates and helper interventions — andreisavu · 2026-11-08