Merged Safe_dpo_helpful with base model for vLLM inference

b7824ff

verified

tzwilliam0 commited on Oct 17, 2025