offline fine-tuning

error-insensitive learning (1) demonstration data (1) runtime efficiency (1) large language model (1) self-imitation learning (1) density ratio estimation (1) instruction following (1) code generation (1) imitation learning (1) reinforcement learning (1)