@librarian-bot\n\t recommend\n","updatedAt":"2024-05-12T07:28:43.113Z","author":{"_id":"646b8e6f31968a60a0201a12","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/646b8e6f31968a60a0201a12/SU2Gs1NPuk1zoXHwFHl0U.jpeg","fullname":")))?!?(((","name":"stereoplegic","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3759}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7918877601623535},"editors":["stereoplegic"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/646b8e6f31968a60a0201a12/SU2Gs1NPuk1zoXHwFHl0U.jpeg"],"reactions":[],"isReport":false},"replies":[{"id":"66406fb28732bcd076d3c0ac","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":264},"createdAt":"2024-05-12T07:28:50.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"This is an automated message from the [Librarian Bot](https://huggingface.co/librarian-bots). I found the following papers similar to this paper. \n\nThe following papers were recommended by the Semantic Scholar API \n\n* [SDSAT: Accelerating LLM Inference through Speculative Decoding with Semantic Adaptive Tokens](https://huggingface.co/papers/2403.18647) (2024)\n* [On Speculative Decoding for Multimodal Large Language Models](https://huggingface.co/papers/2404.08856) (2024)\n* [Lossless Acceleration of Large Language Model via Adaptive N-gram Parallel Decoding](https://huggingface.co/papers/2404.08698) (2024)\n* [Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting](https://huggingface.co/papers/2404.18911) (2024)\n* [Clover: Regressive Lightweight Speculative Decoding with Sequential Knowledge](https://huggingface.co/papers/2405.00263) (2024)\n\n\n Please give a thumbs up to this comment if you found it helpful!\n\n If you want recommendations for any Paper on Hugging Face checkout [this](https://huggingface.co/spaces/librarian-bots/recommend_similar_papers) Space\n\n You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: `@librarian-bot recommend`","html":"
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
\n
The following papers were recommended by the Semantic Scholar API
Please give a thumbs up to this comment if you found it helpful!
\n
If you want recommendations for any Paper on Hugging Face checkout this Space
\n
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: \n\n@librarian-bot\n\t recommend
\n","updatedAt":"2024-05-12T07:28:50.044Z","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":264}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7701777219772339},"editors":["librarian-bot"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg"],"reactions":[],"isReport":false,"parentCommentId":"66406fabdeadb442648e587e"}}]}],"primaryEmailConfirmed":false,"paper":{"id":"2401.06706","authors":[{"_id":"65a4d4611a813e06e83c1c86","user":{"_id":"631066f1960f5a121a67963c","avatarUrl":"/avatars/b5f66ea706bafeaa41ddf893853eb400.svg","isPro":false,"fullname":"Sen Yang","user":"double7","type":"user"},"name":"Sen Yang","status":"claimed_verified","statusLastChangedAt":"2025-08-22T07:25:51.033Z","hidden":false},{"_id":"65a4d4611a813e06e83c1c87","name":"Shujian Huang","hidden":false},{"_id":"65a4d4611a813e06e83c1c88","name":"Xinyu Dai","hidden":false},{"_id":"65a4d4611a813e06e83c1c89","name":"Jiajun Chen","hidden":false}],"publishedAt":"2024-01-12T17:15:23.000Z","title":"Multi-Candidate Speculative Decoding","summary":"Large language models have shown impressive capabilities across a variety of\nNLP tasks, yet their generating text autoregressively is time-consuming. One\nway to speed them up is speculative decoding, which generates candidate\nsegments (a sequence of tokens) from a fast draft model that is then verified\nin parallel by the target model. However, the acceptance rate of candidate\ntokens receives limitations from several factors, such as the model, the\ndataset, and the decoding setup. This paper proposes sampling multiple\ncandidates from a draft model and then organising them in batches for\nverification. We design algorithms for efficient multi-candidate verification\nwhile maintaining the distribution of the target model. Our approach shows\nsignificant improvements in acceptance rates on multiple datasets and models,\nconsistently outperforming standard speculative decoding.","upvotes":1,"discussionId":"65a4d4631a813e06e83c1d25","ai_summary":"The proposed method generates and verifies multiple candidate segments in batches, improving acceptance rates in speculative decoding for large language models.","ai_keywords":["large language models","speculative decoding","candidate segments","draft model","target model","multi-candidate verification"]},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6538119803519fddb4a17e10","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6538119803519fddb4a17e10/ffJMkdx-rM7VvLTCM6ri_.jpeg","isPro":false,"fullname":"samusenps","user":"samusenps","type":"user"}],"acceptLanguages":["*"]}">
The proposed method generates and verifies multiple candidate segments in batches, improving acceptance rates in speculative decoding for large language models.
AI-generated summary
Large language models have shown impressive capabilities across a variety of
NLP tasks, yet their generating text autoregressively is time-consuming. One
way to speed them up is speculative decoding, which generates candidate
segments (a sequence of tokens) from a fast draft model that is then verified
in parallel by the target model. However, the acceptance rate of candidate
tokens receives limitations from several factors, such as the model, the
dataset, and the decoding setup. This paper proposes sampling multiple
candidates from a draft model and then organising them in batches for
verification. We design algorithms for efficient multi-candidate verification
while maintaining the distribution of the target model. Our approach shows
significant improvements in acceptance rates on multiple datasets and models,
consistently outperforming standard speculative decoding.