mirror of
https://github.com/hiyouga/LLaMAFactory
synced 2026-05-12 18:06:33 +02:00
unwrap_model_for_generation(reward_model) is necessary for zero3 training Former-commit-id: c8d5b21700577cae8d6ca03359bcf1762c8b7cb8
unwrap_model_for_generation(reward_model) is necessary for zero3 training Former-commit-id: c8d5b21700577cae8d6ca03359bcf1762c8b7cb8