Top-k Training of GANs: Improving GAN Performance by Throwing Away Bad Samples · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2002.06224/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTTop-k Training of GANs: Improving GAN Performance by Throwing Away Bad Samples — mentioned by gwernhttps://news.ycombinator.com/item?id=42561803urn:hackerpapers:mention:2002.06224:42561803Tue, 31 Dec 2024 21:02:25 GMT<p>I&#x27;m also reminded of top-_k_ training/sampling, where you throw away the inconsistent samples: https://arxiv.org/abs/2002.06224 &quot;Top-k Training of GANs: Improving GAN Performance by Throwing Away Bad Samples&quot;, Sinha et al 2020. Which seems like it would be roughly equivalent to throwing away gradients if they are inconsistent...?</p><p><a href="https://hackerpaper.kaipfister.de/p/2002.06224/">Paper and discussion history</a>. A mention is not an endorsement.</p>stat.ML