Task-level Distributionally Robust Optimization for Large Language Model-based Dense Retrieval

community