ICLR 2026 - Training for Compositional Sensitivity Reduces Dense Retrieval Generalization
This paper was accepted to the SciForDL workshop at the ICLR conference 2026.
Radoslav Ralev, Aditeya Baral, Iliya Zhechev, Jen Agarwal & Srijith Rajamohan
March 25, 2026
This paper was accepted to the SciForDL workshop at the ICLR conference 2026.