Sharing a dataset I built. Disclosure: this is my project. Free to download and use.<\/p>\n
https:\/\/huggingface.co\/datasets\/fineset-io\/model-merging-papers<\/a><\/p>\n Stats:<\/strong><\/p>\n – 381 records, 2021\u20132026<\/p>\n – Sources: arXiv + Semantic Scholar, cross-referenced by arxiv_id and DOI<\/p>\n – quality_score: 0-1, citation-normalized<\/p>\n Fields: id, title, abstract, authors, categories, published_date,<\/p>\n citation_count, quality_score, has_code, code_url, venue<\/p>\n The most-cited paper in the set is “Model soups: averaging weights of multiple<\/p>\n fine-tuned models improves accuracy without increasing inference time” (1,565 citations,<\/p>\n 2022); if you’re doing any merging work this is probably already in your reading list,<\/p>\n but the rest of the dataset has 380 more.<\/p>\n 109 papers have code repos; filter has_code=true if you want reproducible implementations.<\/p>\n Built with FineSet (fineset.io<\/a>). Sign up free to get daily-refreshed datasets on your own topic.<\/p>\n<\/div>\n submitted by \/u\/fineset-io <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
