多模態(tài)獎勵大一統(tǒng)!UNIFIEDREWARD突破任務邊界,圖像視頻雙域性能飆升的秘密
文章鏈接:https:arxiv.orgpdf2503.05236項目鏈接:https:codegoat24.github.ioUnifiedRewardGithub鏈接:https:github.comCodeGoat24UnifiedRewardHuggingface鏈接:https:huggingface.copapers2503.05236Models鏈接:https:huggingface.cocollectionsCodeGoat24unifiedrewardmodels67c3008148c3a380d15ac63aDatasets鏈接:https:huggingface.cocollectionsCodeGoat24unifiedrewardtrainingdata67c300d4fd5eff00fa7f1ede亮點直擊構...