LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches
LoGRA cuts LLM reinforcement-learning memory by up to 45.7% using low-rank gradient sketches.
1 stories, newest first · one card per story
LoGRA cuts LLM reinforcement-learning memory by up to 45.7% using low-rank gradient sketches.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views