1 briefs tagged #lv lm.
The framework uses hard negatives to generate retrieval-focused reasoning for LVLM-based retrievers.