Entrocraft uses rejection sampling to enforce precise entropy schedules in LLM RL by biasing advantages, enabling longer training, better generalization, and higher performance than baselines.
(9) Entropy decreases (∆H<0 ) when the term in parentheses is positive: logp k >log Q i∈V\{k} p − δpi δpk i
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Addressing Performance Saturation for LLM RL via Precise Entropy Curve Control
Entrocraft uses rejection sampling to enforce precise entropy schedules in LLM RL by biasing advantages, enabling longer training, better generalization, and higher performance than baselines.