Paper: Epiplexity Guided Data Selection Improves Generalization
burny_tech · x · 2026-08-17
The paper proposes using Epiplexity as an online training signal for data selection and synthetic data generation. For selection, it fits scaling laws to predict epiplexity gain for adaptive sampling. For generation, it uses REINFORCE to guide the generator. Higher epiplexity correlates with better zero-shot and fine-tuning performance.
Related event: New Metric Epiplexity Redefines Data Information(2 posts)→
More from Research
- Burkov skew AI hype: 'deterministic LLMs' and 'first agents' are old tricks rebranded — burkov · 2026-09-23
- Continuous diffusion beats discrete on random k-SAT, proposed as standard benchmark — ArashVahdat · 2026-09-23
- Grady Booch: Contemporary AI Still Lacks Abductive Reasoning, Just 'Next-Token Prediction' — Grady_Booch · 2026-09-23
- AI solves Navier-Stokes-related problem as machines upend mathematics, New Scientist reports — burny_tech · 2026-09-23
- Mathematician says OpenAI likely proved a significant partial case of the Hodge conjecture — burny_tech · 2026-09-23
- Code benchmarks are mostly slop: dev calls for narrow evals per domain, not one score — almmaasoglu · 2026-09-23