Are We Really Making Much Progress in Text Classification? A Comparative Review · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2204.03954/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTAre We Really Making Much Progress in Text Classification? A Comparative Review — mentioned by PaulHoulehttps://news.ycombinator.com/item?id=35070531urn:hackerpapers:mention:2204.03954:35070531Wed, 08 Mar 2023 15:53:13 GMT<p>I like it! But... I am doing something like this for my intelligent feed reader but using local models instead of the GPT API and using moderate-shot instead of zero-shot. Models like RoBERTa and miniLM frequently can learn pretty good models with a few hundred or thousand examples. I&#x27;d point to https://arxiv.org/pdf/2204.03954.pdf Now for years people thought it would be more scalable for some big co (say Facebook) to train a big model for everybody but that list in there of problems, particularly 2-7 and 10-12 are problems that only exist for the big company and that have to do with…</p><p><a href="https://hackerpaper.kaipfister.de/p/2204.03954/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.CL