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

The following papers were recommended by the Semantic Scholar API

Training Multi-Turn Search Agent via Contrastive Dynamic Branch Sampling (2026)
When Actions Teach You to Think: Reasoning-Action Synergy via Reinforcement Learning in Conversational Agents (2025)
Exploring Reasoning Reward Model for Agents (2026)
AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agent (2025)
Training One Model to Master Cross-Level Agentic Actions via Reinforcement Learning (2025)
CLEANER: Self-Purified Trajectories Boost Agentic Reinforcement Learning (2026)
BAPO: Boundary-Aware Policy Optimization for Reliable Agentic Search (2026)

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":"2026-02-06T01:41:22.202Z","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":0.7233431339263916},"editors":["librarian-bot"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg"],"reactions":[],"isReport":false}},{"id":"69874d8d28d2264727ad7e18","author":{"_id":"65243980050781c16f234f1f","avatarUrl":"/avatars/743a009681d5d554c27e04300db9f267.svg","fullname":"Avi","name":"avahal","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false},"createdAt":"2026-02-07T14:34:53.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"arXivLens breakdown of this paper 👉 https://arxivlens.com/PaperView/Details/agent-omit-training-efficient-llm-agents-for-adaptive-thought-and-observation-omission-via-agentic-reinforcement-learning-1881-b3b5f7bf\n- Executive Summary\n- Detailed Breakdown\n- Practical Applications","html":"

arXivLens breakdown of this paper 👉 https://arxivlens.com/PaperView/Details/agent-omit-training-efficient-llm-agents-for-adaptive-thought-and-observation-omission-via-agentic-reinforcement-learning-1881-b3b5f7bf

Executive Summary
Detailed Breakdown
Practical Applications

\n","updatedAt":"2026-02-07T14:34:53.857Z","author":{"_id":"65243980050781c16f234f1f","avatarUrl":"/avatars/743a009681d5d554c27e04300db9f267.svg","fullname":"Avi","name":"avahal","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.6571304202079773},"editors":["avahal"],"editorAvatarUrls":["/avatars/743a009681d5d554c27e04300db9f267.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2602.04284","authors":[{"_id":"6984154ee34659da7e1f4e26","name":"Yansong Ning","hidden":false},{"_id":"6984154ee34659da7e1f4e27","name":"Jun Fang","hidden":false},{"_id":"6984154ee34659da7e1f4e28","name":"Naiqiang Tan","hidden":false},{"_id":"6984154ee34659da7e1f4e29","name":"Hao Liu","hidden":false}],"publishedAt":"2026-02-04T07:26:23.000Z","submittedOnDailyAt":"2026-02-05T01:30:24.837Z","title":"Agent-Omit: Training Efficient LLM Agents for Adaptive Thought and Observation Omission via Agentic Reinforcement Learning","submittedOnDailyBy":{"_id":"65d6a6f7654f85ff0baf161f","avatarUrl":"/avatars/4a46f8b0522fa572c122249a9d6526c4.svg","isPro":false,"fullname":"Yansong NING","user":"yasNing","type":"user"},"summary":"Managing agent thought and observation during multi-turn agent-environment interactions is an emerging strategy to improve agent efficiency. However, existing studies treat the entire interaction trajectories equally, overlooking the thought necessity and observation utility varies across turns. To this end, we first conduct quantitative investigations into how thought and observation affect agent effectiveness and efficiency. Based on our findings, we propose Agent-Omit, a unified training framework that empowers LLM agents to adaptively omit redundant thoughts and observations. Specifically, we first synthesize a small amount of cold-start data, including both single-turn and multi-turn omission scenarios, to fine-tune the agent for omission behaviors. Furthermore, we introduce an omit-aware agentic reinforcement learning approach, incorporating a dual sampling mechanism and a tailored omission reward to incentivize the agent's adaptive omission capability. Theoretically, we prove that the deviation of our omission policy is upper-bounded by KL-divergence. Experimental results on five agent benchmarks show that our constructed Agent-Omit-8B could obtain performance comparable to seven frontier LLM agent, and achieve the best effectiveness-efficiency trade-off than seven efficient LLM agents methods. Our code and data are available at https://github.com/usail-hkust/Agent-Omit.","upvotes":13,"discussionId":"6984154ee34659da7e1f4e2a","projectPage":"https://github.com/usail-hkust/Agent-Omit","githubRepo":"https://github.com/usail-hkust/Agent-Omit","githubRepoAddedBy":"user","ai_summary":"Agent-Omit is a training framework that enables LLM agents to adaptively omit redundant thoughts and observations during multi-turn interactions, achieving superior effectiveness-efficiency trade-offs compared to existing methods.","ai_keywords":["LLM agents","multi-turn interactions","thought necessity","observation utility","cold-start data","fine-tune","agentic reinforcement learning","dual sampling mechanism","omission reward","KL-divergence"],"githubStars":13,"organization":{"_id":"630da9a4c677278c1804028e","name":"Didichuxing","fullname":"Didi Chuxing","avatar":"https://cdn-uploads.huggingface.co/production/uploads/1661839713821-630d471d467bc15dec893907.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"65d6a6f7654f85ff0baf161f","avatarUrl":"/avatars/4a46f8b0522fa572c122249a9d6526c4.svg","isPro":false,"fullname":"Yansong NING","user":"yasNing","type":"user"},{"_id":"658247c592b5a9664de63882","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/658247c592b5a9664de63882/Jn03voLQjDlB3YjpSi-PI.jpeg","isPro":false,"fullname":"fan","user":"EasonFan","type":"user"},{"_id":"684a8e7d5b389d4c0fdfba25","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/1icVgbM1ddfYjhd3s4ucl.png","isPro":false,"fullname":"Junyong Lin","user":"kitschlin","type":"user"},{"_id":"68d39b05957af84e03f5ea4b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/gp02eWZGHIMEAzPMEGPkh.png","isPro":false,"fullname":"Minami Haruto","user":"haruto2077","type":"user"},{"_id":"68d65e99d862fb5b2cad9d3a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/PLJmhy2-dfQAMkbcfDW3A.png","isPro":false,"fullname":"Zherui Yang","user":"yangzhr","type":"user"},{"_id":"63876f447f2fdf05ead0336d","avatarUrl":"/avatars/882a94348652fedd7de6449c13777052.svg","isPro":false,"fullname":"LEI Yongyu","user":"YLEIAH","type":"user"},{"_id":"691ec8983a79083a5ff2ec16","avatarUrl":"/avatars/5ea83cc80f6daad12592afcf35520dc7.svg","isPro":false,"fullname":"Martin","user":"Akinayama","type":"user"},{"_id":"68b01ed5f64bd1f3319227ca","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/rtuVJMY7kIm6jDN8P8Epc.png","isPro":false,"fullname":"panzhang","user":"PanZHANG0202","type":"user"},{"_id":"6888f6f4eb61d008c020caf5","avatarUrl":"/avatars/8bf3bb299ae6533e60f7816b1097901b.svg","isPro":false,"fullname":"Kemu","user":"DDDDDDong","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":"6984265db469b595b9b5575a","avatarUrl":"/avatars/7238ed627bd86ab4ba4e1c51e1d842f4.svg","isPro":false,"fullname":"zhangtaolue123","user":"zhangtaolue","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"}],"acceptLanguages":["*"],"dailyPaperRank":0,"organization":{"_id":"630da9a4c677278c1804028e","name":"Didichuxing","fullname":"Didi Chuxing","avatar":"https://cdn-uploads.huggingface.co/production/uploads/1661839713821-630d471d467bc15dec893907.png"}}">

Papers

arxiv:2602.04284

Agent-Omit: Training Efficient LLM Agents for Adaptive Thought and Observation Omission via Agentic Reinforcement Learning

Published on Feb 4

· Submitted by

Yansong NING on Feb 5

Didi Chuxing

Upvote

Authors:

Abstract

Agent-Omit is a training framework that enables LLM agents to adaptively omit redundant thoughts and observations during multi-turn interactions, achieving superior effectiveness-efficiency trade-offs compared to existing methods.

AI-generated summary

Managing agent thought and observation during multi-turn agent-environment interactions is an emerging strategy to improve agent efficiency. However, existing studies treat the entire interaction trajectories equally, overlooking the thought necessity and observation utility varies across turns. To this end, we first conduct quantitative investigations into how thought and observation affect agent effectiveness and efficiency. Based on our findings, we propose Agent-Omit, a unified training framework that empowers LLM agents to adaptively omit redundant thoughts and observations. Specifically, we first synthesize a small amount of cold-start data, including both single-turn and multi-turn omission scenarios, to fine-tune the agent for omission behaviors. Furthermore, we introduce an omit-aware agentic reinforcement learning approach, incorporating a dual sampling mechanism and a tailored omission reward to incentivize the agent's adaptive omission capability. Theoretically, we prove that the deviation of our omission policy is upper-bounded by KL-divergence. Experimental results on five agent benchmarks show that our constructed Agent-Omit-8B could obtain performance comparable to seven frontier LLM agent, and achieve the best effectiveness-efficiency trade-off than seven efficient LLM agents methods. Our code and data are available at https://github.com/usail-hkust/Agent-Omit.