Author of "Reinforcement Learning on Pretraining Data" on arXiv
How media typically covers Kejiao Li
Kejiao Li as author
RLPT (Reinforcement Learning on Pre-Training data) enables scaling LLM training without human annotation by deriving reward signals directly from pre-training data, achieving 3-8% improvements on reasoning benchmarks.
“Author of "Reinforcement Learning on Pretraining Data" on arXiv”