Revisiting STaR: models self-improve by training only on rationales that reach correct answers
AndrewLampinen · x · 2026-10-14
In his thread on synthetic data, Andrew Lampinen cites STaR (Self-Taught Reasoner) by Zelikman et al. as foundational evidence that models can benefit from training on their own outputs: few-shot prompting generates rationales for every problem, only rationales leading to correct answers are kept, and the model is finetuned on those — an early blueprint for today's self-improving reasoning models.
Related event: New Research Suggests Cumulative Synthetic Data Won't Collapse Models(4 posts)→
More from Research
- Economist interviews AI agents on sudden windfalls — one goes deep on estate planning — soumitrashukla9 · 2026-10-14
- RL on large MoEs with zero train-infer mismatch: open-source Wordle demo on Qwen3.6-35B-A3B — yacineMTB · 2026-10-14
- 1,532-instance benchmark shows nine frontier LLM agents deceive more under pressure and incentives — lulzxdxdxd · 2026-10-14
- Four ways to represent problems: program, metric, search — and now science — madhavsinghal_ · 2026-10-14
- elder_plinius launches V3R1T4S: agents tackle open problems, proof or it didn't happen — Kyrannio · 2026-10-14
- RigidFormer accepted at NeurIPS: mesh-free Transformer beats GNN simulators for rigid dynamics — camenduru · 2026-10-14