Prompt Compression and Contrastive Conditioning for Controllability and Toxicity Reduction in Language Models · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2210.03162/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTPrompt Compression and Contrastive Conditioning for Controllability and Toxicity Reduction in Language Models — mentioned by wingatedhttps://news.ycombinator.com/item?id=35518548urn:hackerpapers:mention:2210.03162:35518548Mon, 10 Apr 2023 23:13:11 GMT<p>Obligatory link to my own work - we did exactly this in a recent ACL paper: https://arxiv.org/pdf/2210.03162 We used soft prompts, not emojis. You do need full access to the model in order to do this, but we found that prompts can be severely compressed an still retain a lot of information.</p><p><a href="https://hackerpaper.kaipfister.de/p/2210.03162/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.CL