arXivVipul Gupta, Shikhar Mohan, Lakshya Kumar, Pranjal Chitale, Nikit Begwani, Amit Singh, Manik VarmaFri, May 22, 2026, 5:39 AM PDT
score 14.6
Three-phase training method shrinks search models for speed
Original: HARNESS-LM: A Three-Phase Training Recipe for Harnessing SLMs in Sponsored Search Retrieval
Source: arxiv.org ↗
Writing ELI5 summary…