{"id":27123697,"url":"https://github.com/Eclipsess/Awesome-Efficient-Reasoning-LLMs","last_synced_at":"2025-04-07T13:01:51.633Z","repository":{"id":283589407,"uuid":"951125019","full_name":"Eclipsess/Awesome-Efficient-Reasoning-LLMs","owner":"Eclipsess","description":"A Survey on Efficient Reasoning for LLMs","archived":false,"fork":false,"pushed_at":"2025-04-01T18:19:35.000Z","size":1251,"stargazers_count":230,"open_issues_count":2,"forks_count":5,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-04-01T19:32:08.899Z","etag":null,"topics":["efficiency","large-language-models","large-reasoning-models"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Eclipsess.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2025-03-19T07:43:22.000Z","updated_at":"2025-04-01T18:19:38.000Z","dependencies_parsed_at":"2025-03-21T03:24:29.738Z","dependency_job_id":"b11e9be2-6be8-4d5d-961b-4af5f86f0418","html_url":"https://github.com/Eclipsess/Awesome-Efficient-Reasoning-LLMs","commit_stats":null,"previous_names":["eclipsess/awesome-efficient-reasoning-llms"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Eclipsess%2FAwesome-Efficient-Reasoning-LLMs","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Eclipsess%2FAwesome-Efficient-Reasoning-LLMs/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Eclipsess%2FAwesome-Efficient-Reasoning-LLMs/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Eclipsess%2FAwesome-Efficient-Reasoning-LLMs/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Eclipsess","download_url":"https://codeload.github.com/Eclipsess/Awesome-Efficient-Reasoning-LLMs/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247657273,"owners_count":20974344,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["efficiency","large-language-models","large-reasoning-models"],"created_at":"2025-04-07T13:01:46.172Z","updated_at":"2025-04-07T13:01:51.590Z","avatar_url":"https://github.com/Eclipsess.png","language":null,"funding_links":[],"categories":["A01_文本生成_文本对话","Related Survey","Resources","Other Lists","Topics"],"sub_categories":["大语言对话模型及数据","Efficient Reasoning","🔖 Future Directions","TeX Lists","Applications","LLM Reasoning"],"readme":"# Awesome-Efficient-Reasoning-LLMs\n\n[![arXiv](https://img.shields.io/badge/arXiv-Stop_Overthinking-b31b1b.svg)](https://arxiv.org/abs/2503.16419)\n\u003c!-- [![Maintenance](https://img.shields.io/badge/Maintained%3F-yes-green.svg)]() \u003c!-- Optional: Link to GitHub repo --\u003e\n\u003c!-- [![Last Commit](https://img.shields.io/github/last-commit/\u003cyour-username\u003e/\u003crepo-name\u003e)]() \u003c!-- Fill in your repo link --\u003e\n\u003c!-- [![Contributions Welcome](https://img.shields.io/badge/Contributions-welcome-blue)]() --\u003e \n\n\u003c!-- omit in toc --\u003e\n## 📢 News\n- **March 20, 2025**: We release the first survey for efficient reasoning of LLMs \"[Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models](https://arxiv.org/abs/2503.16419)\".  \n  Feel free to cite, contribute, or open a pull request to add recent related papers!\n\n\u003c!-- omit in toc --\u003e\n![Pipeline](./figs/pipeline1.png)\n\nIn this paper, we present the first structured survey that systematically investigates and organizes the current progress in achieving **efficient reasoning in LLMs**.\n\n## 📊 Taxonomy\n\nBelow is a taxonomy graph summarizing the current landscape of efficient reasoning research for LLMs:\n\n![Taxonomy](./figs/taxonomy.png)\n\n---\n\n\u003c!-- omit in toc --\u003e\n## 📚 Table of Contents\n\n- [Awesome-Efficient-Reasoning-LLM](#awesome-efficient-reasoning-llm)\n  - **Model-based Efficient Reasoning**\n    - [Section I: RL with Length Reward Design](#section-i-rl-with-length-reward-design)\n    - [Section II: SFT with Variable-Length CoT Data](#section-ii-sft-with-variable-length-cot-data)\n  - **Reasoning Output-based Efficient Reasoning**\n    - [Section III: Compressing Reasoning Steps into Fewer Latent Representation](#section-iii-compressing-reasoning-steps-into-fewer-latent-representation)\n    - [Section IV: Dynamic Reasoning Paradigm during Inference](#section-iv-dynamic-reasoning-paradigm-during-inference)\n  - **Input Prompt-based Efficient Reasoning**\n    - [Section V: Prompt-Guided Efficient Reasoning](#section-v-prompt-guided-efficient-reasoning)\n    - [Section VI: Prompts Attribute-Driven Reasoning Routing](#section-vi-prompts-attribute-driven-reasoning-routing)\n  - **Reasoning Abilities with Efficient Data and Small Language Models**\n    - [Section VII: Reasoning Abilities via Efficient Training Data and Model Compression](#section-vii-reasoning-abilities-via-efficient-training-data-and-model-compression)\n  - **Evaluation and Benchmark**\n    - [Section VIII: Evaluation and Benchmark](#section-viii-evaluation-and-benchmark)\n\n\n---\n\n\u003c!--[[Paper]](pdf LINK) ![](https://img.shields.io/badge/pdf-\u003c TIME \u003e-red)--\u003e\n\n## Section I:  RL with Length Reward Design\n\n* Demystifying Long Chain-of-Thought Reasoning in LLMs [[Paper]](https://arxiv.org/pdf/2502.03373) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning [[Paper]](https://arxiv.org/pdf/2501.12570) ![](https://img.shields.io/badge/pdf-2025.01-red)\n* Kimi k1.5: Scaling Reinforcement Learning with LLMs [[Paper]](https://arxiv.org/pdf/2501.12599) ![](https://img.shields.io/badge/pdf-2025.01-red)\n* Training Language Models to Reason Efficiently [[Paper]](https://arxiv.org/pdf/2502.04463) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning [[Paper]](https://www.arxiv.org/pdf/2503.04697) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models [[Paper]](https://arxiv.org/pdf/2503.04472) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* Optimizing Test-Time Compute via Meta Reinforcement Fine-Tuning [[Paper]](https://arxiv.org/pdf/2503.07572) ![](https://img.shields.io/badge/pdf-2025.03-red)\n\n## Section II: SFT with Variable-Length CoT Data\n\n* TokenSkip: Controllable Chain-of-Thought Compression in LLMs [[Paper]](https://arxiv.org/pdf/2502.12067) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* C3oT: Generating Shorter Chain-of-Thought without Compromising Effectiveness [[Paper]](https://arxiv.org/pdf/2412.11664) ![](https://img.shields.io/badge/pdf-2024.12-red)\n* CoT-Valve: Length-Compressible Chain-of-Thought Tuning [[Paper]](https://arxiv.org/pdf/2502.09601) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Self-Training Elicits Concise Reasoning in Large Language Models [[Paper]](https://arxiv.org/pdf/2502.20122) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Distilling System 2 into System 1 [[Paper]](https://arxiv.org/pdf/2407.06023) ![](https://img.shields.io/badge/pdf-2024.07-red)\n* Can Language Models Learn to Skip Steps? [[Paper]](https://arxiv.org/pdf/2411.01855) ![](https://img.shields.io/badge/pdf-2024.11-red)\n* Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Models [[Paper]](https://arxiv.org/pdf/2502.13260) ![](https://img.shields.io/badge/pdf-2025.02-red)\n\n## Section III: Compressing Reasoning Steps into Fewer Latent Representation\n\n* Training Large Language Models to Reason in a Continuous Latent Space [[Paper]](https://arxiv.org/pdf/2412.06769) ![](https://img.shields.io/badge/pdf-2024.12-red)\n* Compressed Chain of Thought: Efficient Reasoning through Dense Representations [[Paper]](https://arxiv.org/pdf/2412.13171) ![](https://img.shields.io/badge/pdf-2024.12-red)\n* Efficient Reasoning with Hidden Thinking (MLLM) [[Paper]](https://arxiv.org/pdf/2501.19201) ![](https://img.shields.io/badge/pdf-2025.01-red)\n* SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMs [[Paper]](https://arxiv.org/pdf/2502.12134) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning [[Paper]](https://arxiv.org/pdf/2502.03275) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Reasoning with Latent Thoughts: On the Power of Looped Transformers [[Paper]](https://arxiv.org/pdf/2502.17416) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation [[Paper]](https://arxiv.org/pdf/2502.21074) ![](https://img.shields.io/badge/pdf-2025.02-red)\n\n## Section IV: Dynamic Reasoning Paradigm during Inference\n\n* Efficiently Serving LLM Reasoning Programs with Certaindex [[Paper]](https://arxiv.org/pdf/2412.20993) ![](https://img.shields.io/badge/pdf-2024.12-red)\n* When More is Less: Understanding Chain-of-Thought Length in LLMs [[Paper]](https://arxiv.org/pdf/2502.07266) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching [[Paper]](https://arxiv.org/pdf/2503.05179) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* Reward-Guided Speculative Decoding for Efficient LLM Reasoning [[Paper]](https://arxiv.org/pdf/2501.19324) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Fast Best-of-N Decoding via Speculative Rejection [[Paper]](https://arxiv.org/pdf/2410.20290) ![](https://img.shields.io/badge/pdf-2024.10-red)\n* FastMCTS: A Simple Sampling Strategy for Data Synthesis [[Paper]](https://www.arxiv.org/pdf/2502.11476) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Dynamic Parallel Tree Search for Efficient LLM Reasoning [[Paper]](https://arxiv.org/pdf/2502.16235) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Sampling-Efficient Test-Time Scaling: Self-Estimating the Best-of-N Sampling in Early Decoding [[Paper]](https://arxiv.org/pdf/2503.01422) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* LightThinker: Thinking Step-by-Step Compression (training LLMs to compress thoughts into gist tokens) [[Paper]](https://arxiv.org/pdf/2502.15589) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* InftyThink: Breaking the Length Limits of Long-Context Reasoning in Large Language Models [[Paper]](https://www.arxiv.org/pdf/2503.06692) ![](https://img.shields.io/badge/pdf-2025.03-red)\n  \n## Section V: Prompt-Guided Efficient Reasoning\n\n* Token-Budget-Aware LLM Reasoning [[Paper]](https://arxiv.org/pdf/2412.18547) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Chain of Draft: Thinking Faster by Writing Less [[Paper]](https://arxiv.org/pdf/2502.18600) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* How Well do LLMs Compress Their Own Chain-of-Thought? A Token Complexity Approach [[Paper]](https://arxiv.org/pdf/2503.01141) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* The Benefits of a Concise Chain of Thought on Problem-Solving in Large Language Models [[Paper]](https://arxiv.org/pdf/2401.05618) ![](https://img.shields.io/badge/pdf-2024.10-red)\n\n## Section VI: Prompts Attribute-Driven Reasoning Routing\n* Claude 3.7 Sonnet and Claude Code [[website]](https://www.anthropic.com/news/claude-3-7-sonnet) ![](https://img.shields.io/badge/html-2025.02-red)\n* Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching [[Paper]](https://arxiv.org/pdf/2503.05179) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* Learning to Route LLMs with Confidence Tokens [[Paper]](https://arxiv.org/pdf/2410.13284) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Confident or Seek Stronger: Exploring Uncertainty-Based On-device LLM Routing From Benchmarking to Generalization [[Paper]](https://arxiv.org/pdf/2502.04428) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* RouteLLM: Learning to Route LLMs with Preference Data [[Paper]](https://arxiv.org/pdf/2406.18665) ![](https://img.shields.io/badge/pdf-2025.02-red)\n\n## Section VII: Reasoning Abilities via Efficient Training Data and Model Compression\n\n* LIMO: Less is More for Reasoning [[Paper]](https://arxiv.org/pdf/2502.03387) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* s1: Simple test-time scaling [[Paper]](https://arxiv.org/pdf/2501.19393) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* S2R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learning [[Paper]](https://arxiv.org/pdf/2502.12853) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyond [[Paper]](https://arxiv.org/pdf/2503.10460) ![](https://img.shields.io/badge/pdf-2025.03-red)\n* Small Models Struggle to Learn from Strong Reasoners [[Paper]](https://arxiv.org/pdf/2502.12143) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Towards Reasoning Ability of Small Language Models [[Paper]](https://arxiv.org/pdf/2502.11569) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Mixed Distillation Helps Smaller Language Models Reason Better [[Paper]](https://arxiv.org/pdf/2312.10730) ![](https://img.shields.io/badge/pdf-2024.02-red)\n* Small language models need strong verifiers to self-correct reasoning [[Paper]](https://arxiv.org/pdf/2404.17140) ![](https://img.shields.io/badge/pdf-2024.06-red)\n* Teaching Small Language Models Reasoning through Counterfactual Distillation [[Paper]](https://aclanthology.org/2024.emnlp-main.333.pdf) ![](https://img.shields.io/badge/pdf-2024.11-red)\n* Improving Mathematical Reasoning Capabilities of Small Language Models via Feedback-Driven Distillation [[Paper]](https://arxiv.org/pdf/2411.14698) ![](https://img.shields.io/badge/pdf-2024.11-red)\n* Probe then retrieve and reason: Distilling probing and reasoning capabilities into smaller language models [[Paper]](https://aclanthology.org/2024.lrec-main.1140.pdf) ![](https://img.shields.io/badge/pdf-2024.05-red)\n* Distilling Reasoning Ability from Large Language Models with Adaptive Thinking [[Paper]](https://arxiv.org/pdf/2404.09170) ![](https://img.shields.io/badge/pdf-2024.08-red)\n* SKIntern: Internalizing Symbolic Knowledge for Distilling Better CoT Capabilities into Small Language Models [[Paper]](https://arxiv.org/pdf/2409.13183) ![](https://img.shields.io/badge/pdf-2024.12-red)\n  \n## Section VIII: Evaluation and Benchmark\n* Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling [[Paper]](https://arxiv.org/pdf/2502.06703) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* The Danger of Overthinking: Examining the Reasoning-Action Dilemma in Agentic Tasks [[Paper]](https://arxiv.org/pdf/2502.08235) ![](https://img.shields.io/badge/pdf-2025.02-red)\n* Inference-Time Computations for LLM Reasoning and Planning: A Benchmark and Insights [[Paper]](https://arxiv.org/pdf/2502.12521) ![](https://img.shields.io/badge/pdf-2025.02-red)\n\n\n\n\n## Citation\nIf you find this work useful, welcome to cite us.\n```bib\n@article{sui2025stop,\n  title={Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models},\n  author={Sui, Yang and Chuang, Yu-Neng and Wang, Guanchu and Zhang, Jiamu and Zhang, Tianyi and Yuan, Jiayi and Liu, Hongyi and Wen, Andrew and Chen, Hanjie and Hu, Xia and others},\n  journal={arXiv preprint arXiv:2503.16419},\n  year={2025}\n}\n```\n\n## Acknowledgment\n\u003e 🧩 *Layout inspired by [zzli2022/Awesome-System2-Reasoning-LLM](https://github.com/zzli2022/Awesome-System2-Reasoning-LLM). Many thanks for the great structure!*\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FEclipsess%2FAwesome-Efficient-Reasoning-LLMs","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FEclipsess%2FAwesome-Efficient-Reasoning-LLMs","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FEclipsess%2FAwesome-Efficient-Reasoning-LLMs/lists"}