Librarian Bot. I found the following papers similar to this paper.

The following papers were recommended by the Semantic Scholar API

Rethinking Predictive Modeling for LLM Routing: When Simple kNN Beats Complex Learned Routers (2025)
Router-R1: Teaching LLMs Multi-Round Routing and Aggregation via Reinforcement Learning (2025)
Towards Efficient Multi-LLM Inference: Characterization and Analysis of LLM Routing and Hierarchical Techniques (2025)
INFERENCEDYNAMICS: Efficient Routing Across LLMs through Structured Capability and Knowledge Profiling (2025)
Hybrid AI for Responsive Multi-Turn Online Conversations with Novel Dynamic Routing and Feedback Adaptation (2025)
TagRouter: Learning Route to LLMs through Tags for Open-Domain Text Generation Tasks (2025)
Query Routing for Retrieval-Augmented Language Models (2025)

Please give a thumbs up to this comment if you found it helpful!

If you want recommendations for any Paper on Hugging Face checkout this Space

You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: \n\n@librarian-bot\n\t recommend

\n","updatedAt":"2025-06-28T01:34:53.012Z","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":318,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7548947930335999},"editors":["librarian-bot"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2506.16655","authors":[{"_id":"6858de6bc0c8e29df8ea3d03","user":{"_id":"64650231611ae99d14d30c5b","avatarUrl":"/avatars/1f323e5342e007c7f962ca12ce516102.svg","isPro":false,"fullname":"Co Tran","user":"cotran2","type":"user"},"name":"Co Tran","status":"admin_assigned","statusLastChangedAt":"2025-09-11T09:53:32.552Z","hidden":false},{"_id":"6858de6bc0c8e29df8ea3d04","user":{"_id":"66b681906c8d3b36786b764c","avatarUrl":"/avatars/c04e97278b2275e2b34182229efa1c20.svg","isPro":true,"fullname":"Salman","user":"parachas","type":"user"},"name":"Salman Paracha","status":"claimed_verified","statusLastChangedAt":"2025-06-23T08:14:15.659Z","hidden":false},{"_id":"6858de6bc0c8e29df8ea3d05","user":{"_id":"651e049a855ec8ba9f291154","avatarUrl":"/avatars/db509ed4c1f96e5ff7758b14d1fe46a5.svg","isPro":false,"fullname":"Adil Hafeez","user":"adilhafeez","type":"user"},"name":"Adil Hafeez","status":"admin_assigned","statusLastChangedAt":"2025-09-11T09:52:10.972Z","hidden":false},{"_id":"6858de6bc0c8e29df8ea3d06","user":{"_id":"622e9e56165ba2c1bcbc76da","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1648869498279-622e9e56165ba2c1bcbc76da.jpeg","isPro":false,"fullname":"Shuguang Chen","user":"nehcgs","type":"user"},"name":"Shuguang Chen","status":"claimed_verified","statusLastChangedAt":"2025-06-24T08:09:48.654Z","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/66b681906c8d3b36786b764c/yAiHzS_Ymy5geNcfh2s1K.png"],"publishedAt":"2025-06-19T23:57:41.000Z","submittedOnDailyAt":"2025-06-27T03:00:21.170Z","title":"Arch-Router: Aligning LLM Routing with Human Preferences","submittedOnDailyBy":{"_id":"66b681906c8d3b36786b764c","avatarUrl":"/avatars/c04e97278b2275e2b34182229efa1c20.svg","isPro":true,"fullname":"Salman","user":"parachas","type":"user"},"summary":"With the rapid proliferation of large language models (LLMs) -- each\noptimized for different strengths, style, or latency/cost profile -- routing\nhas become an essential technique to operationalize the use of different\nmodels. However, existing LLM routing approaches are limited in two key ways:\nthey evaluate performance using benchmarks that often fail to capture human\npreferences driven by subjective evaluation criteria, and they typically select\nfrom a limited pool of models. In this work, we propose a preference-aligned\nrouting framework that guides model selection by matching queries to\nuser-defined domains (e.g., travel) or action types (e.g., image editing) --\noffering a practical mechanism to encode preferences in routing decisions.\nSpecifically, we introduce Arch-Router, a compact 1.5B model that\nlearns to map queries to domain-action preferences for model routing decisions.\nOur approach also supports seamlessly adding new models for routing without\nrequiring retraining or architectural modifications. Experiments on\nconversational datasets demonstrate that our approach achieves state-of-the-art\n(SOTA) results in matching queries with human preferences, outperforming top\nproprietary models. Our approach captures subjective evaluation criteria and\nmakes routing decisions more transparent and flexible. Our model is available\nat: https://huggingface.co/katanemo/Arch-Router-1.5B.","upvotes":17,"discussionId":"6858de6bc0c8e29df8ea3d07","projectPage":"https://huggingface.co/katanemo/Arch-Router-1.5B","githubRepo":"https://github.com/katanemo/archgw","githubRepoAddedBy":"user","ai_summary":"A preference-aligned routing framework using a compact 1.5B model effectively matches queries to user-defined domains and action types, outperforming proprietary models in subjective evaluation criteria.","ai_keywords":["large language models","LLM routing","Arch-Router","domain-action preferences"],"githubStars":5117},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"66b681906c8d3b36786b764c","avatarUrl":"/avatars/c04e97278b2275e2b34182229efa1c20.svg","isPro":true,"fullname":"Salman","user":"parachas","type":"user"},{"_id":"651e049a855ec8ba9f291154","avatarUrl":"/avatars/db509ed4c1f96e5ff7758b14d1fe46a5.svg","isPro":false,"fullname":"Adil Hafeez","user":"adilhafeez","type":"user"},{"_id":"6751ae913f59c62f77583757","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/_dCyD7vTFPlFETR97WyQj.png","isPro":false,"fullname":"Ivy Zhang","user":"Ivy1997","type":"user"},{"_id":"6270324ebecab9e2dcf245de","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6270324ebecab9e2dcf245de/cMbtWSasyNlYc9hvsEEzt.jpeg","isPro":false,"fullname":"Kye Gomez","user":"kye","type":"user"},{"_id":"64650231611ae99d14d30c5b","avatarUrl":"/avatars/1f323e5342e007c7f962ca12ce516102.svg","isPro":false,"fullname":"Co Tran","user":"cotran2","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"65dba1f1b62d242ed88b2d2a","avatarUrl":"/avatars/e35ef7687e217e6ab71ad76cef59ea21.svg","isPro":false,"fullname":"Gibran Iqbal","user":"Jibbscript","type":"user"},{"_id":"658d8f9c428114945448026b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/658d8f9c428114945448026b/Sc0YQODIt-u2KbyVZhcX4.jpeg","isPro":false,"fullname":"Rix Beck","user":"rixbee","type":"user"},{"_id":"663ccbff3a74a20189d4aa2e","avatarUrl":"/avatars/83a54455e0157480f65c498cd9057cf2.svg","isPro":false,"fullname":"Nguyen Van Thanh","user":"NguyenVanThanhHust","type":"user"},{"_id":"6387658188b39a64e1eb42a8","avatarUrl":"/avatars/f6bc42e244750eb1acb0be7659e604b3.svg","isPro":false,"fullname":"Qian Wu","user":"Fivethousand","type":"user"},{"_id":"62d32375c72c791b238cd586","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/62d32375c72c791b238cd586/d9NTrsherhh_4Vy-vHF2e.jpeg","isPro":false,"fullname":"julien perez","user":"jnm38","type":"user"},{"_id":"646def60df618b303b419323","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/646def60df618b303b419323/JLJGYen4-5M8ivsLsSk0w.jpeg","isPro":false,"fullname":"Lei Wang","user":"demolei","type":"user"}],"acceptLanguages":["*"],"dailyPaperRank":0}">

Papers

arxiv:2506.16655

Arch-Router: Aligning LLM Routing with Human Preferences

Published on Jun 19, 2025

· Submitted by

Salman on Jun 27, 2025

Upvote

Authors:

Co Tran ,

Salman Paracha ,

Adil Hafeez ,

Shuguang Chen

Abstract

A preference-aligned routing framework using a compact 1.5B model effectively matches queries to user-defined domains and action types, outperforming proprietary models in subjective evaluation criteria.

AI-generated summary

With the rapid proliferation of large language models (LLMs) -- each optimized for different strengths, style, or latency/cost profile -- routing has become an essential technique to operationalize the use of different models. However, existing LLM routing approaches are limited in two key ways: they evaluate performance using benchmarks that often fail to capture human preferences driven by subjective evaluation criteria, and they typically select from a limited pool of models. In this work, we propose a preference-aligned routing framework that guides model selection by matching queries to user-defined domains (e.g., travel) or action types (e.g., image editing) -- offering a practical mechanism to encode preferences in routing decisions. Specifically, we introduce Arch-Router, a compact 1.5B model that learns to map queries to domain-action preferences for model routing decisions. Our approach also supports seamlessly adding new models for routing without requiring retraining or architectural modifications. Experiments on conversational datasets demonstrate that our approach achieves state-of-the-art (SOTA) results in matching queries with human preferences, outperforming top proprietary models. Our approach captures subjective evaluation criteria and makes routing decisions more transparent and flexible. Our model is available at: https://huggingface.co/katanemo/Arch-Router-1.5B.

View arXiv page View PDF Project page GitHub 5.12k Add to collection

Community

parachas

Paper author Paper submitter Jun 27, 2025

This paper discusses a preference-aligned routing framework for LLMs that guides model selection by matching queries to user-defined domains (e.g., travel) or action types (e.g., image editing) -- offering a practical mechanism to encode preferences in routing decisions. Specifically, we introduce Arch-Router, a compact 1.5B model that learns to map queries to domain-action preferences for model routing decisions.