Breaking MLPerf Training: A Case Study on Optimizing BERT · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2402.02447/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTBreaking MLPerf Training: A Case Study on Optimizing BERT — mentioned by Izikiel43https://news.ycombinator.com/item?id=45689884urn:hackerpapers:mention:2402.02447:45689884Fri, 24 Oct 2025 01:56:24 GMT<p>Can you confirm if this generated code is the same as https://arxiv.org/pdf/2402.02447 ?</p><p><a href="https://hackerpaper.kaipfister.de/p/2402.02447/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.LGBreaking MLPerf Training: A Case Study on Optimizing BERT — mentioned by bgwalterhttps://news.ycombinator.com/item?id=45689663urn:hackerpapers:mention:2402.02447:45689663Fri, 24 Oct 2025 01:24:45 GMT<p>I&#x27;m not sure if this is the exact same thing, but a load balancing paper reported a 4.2x speedup by applying a &quot;snake pattern&quot; in 2024: https://arxiv.org/pdf/2402.02447</p><p><a href="https://hackerpaper.kaipfister.de/p/2402.02447/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.LG