knapsack-rl-budget-allocation

【效率与协作】Improve RL training for LLMs by dynamically allocating exploration budget (rollout count) to tasks based on their difficulty and current learning status. Solves the knapsack problem of maximizing gradient signal within fixed compute budget, increasing non-zero policy gradients by 20-40% and achieving 2-4 point performance gains.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准