3 ms·Training a SOTA 7B LRM with 1.5x Throughput Improvment2 points by jinqueeny 2y agodeleted 2y ago[deleted]