mlsanityData to 7 Oct 2026

LIMA: Less Is More for Alignment

arXiv 2305.11206 ↗First posted 18 May 2023cs.CL cs.AI cs.LGcited by 1,080

On the map: LLM alignment / RLHF / preference optimisation → LLM preference optimization methods

Abstract (arXiv)

Large language models are trained in two stages: (1) unsupervised pretraining from raw text, to learn general-purpose representations, and (2) large scale instruction tuning and reinforcement learning, to better align to end tasks and user preferences. We measure the relative importance of these two stages by training LIMA, a 65B parameter LLaMa language model fine-tuned with the standard supervised loss on only 1,000 carefully curated prompts and responses, without any reinforcement learning or human preference modeling. LIMA demonstrates remarkably strong performance, learning to follow specific response formats from only a handful of examples in the training data, including complex queries that range from planning trip itineraries to speculating about alternate history. Moreover, the model tends to generalize well to unseen tasks that did not appear in the training data. In a controlled human study, responses from LIMA are either equivalent or strictly preferred to GPT-4 in 43% of cases; this statistic is as high as 58% when compared to Bard and 65% versus DaVinci003, which was trained with human feedback. Taken together, these results strongly suggest that almost all knowledge in large language models is learned during pretraining, and only limited instruction tuning data is necessary to teach models to produce high quality output.

Key facts. Cited by 1,080 papers in the map (data through 7 Oct 2026). Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 20 of its 25 references that are in the map. First posted on arXiv 18 May 2023.

Builds on

20 of 25 references.

Closest ancestors · 4

Same direction, newest first

  1. 2023Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision4 May
  2. 2023OpenAssistant Conversations -- Democratizing Large Language Model Alignment14 Apr
  3. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructions20 Dec
  4. 2022Training language models to follow instructions with human feedback4 Mar

Other references · 16

  1. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  3. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  4. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  5. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,747
  6. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  7. 2022Scaling Instruction-Finetuned Language Modelscited by 2,997
  8. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  9. 2019The Curious Case of Neural Text Degenerationcited by 2,881
  10. 2021Multitask Prompted Training Enables Zero-Shot Task Generalizationcited by 1,462
  11. 2022Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Taskscited by 858
  12. 2019CTRL: A Conditional Transformer Language Model for Controllable Generationcited by 754
4 more
  1. 2022Unnatural Instructions: Tuning Language Models with (Almost) No Human Laborcited by 395
  2. 2020The Pushshift Reddit Datasetcited by 289
  3. 2020The Turking Test: Can Language Models Understand Instructions?cited by 82
  4. 2021A Few More Examples May Be Worth Billions of Parameterscited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 191; 2024: 436; 2025: 291; 2026: 162.

1,080 papers.

Sort:

By area:

  1. 2023Improved Baselines with Visual Instruction Tuningcited by 4,362
  2. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,582
  3. 2025s1: Simple test-time scalingcited by 1,311
  4. 2023Large Language Models are not Fair Evaluatorscited by 1,050
  5. 2023Enhancing Chat Language Models by Scaling High-quality Instructional Conversationscited by 825
  6. 2023Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedbackcited by 743
  7. 2023MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Modelscited by 724
  8. 2023Aligning Large Multimodal Models with Factually Augmented RLHFcited by 703
  9. 2023Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Modelscited by 695
  10. 2024A Survey on Large Language Models for Code Generationcited by 646
  11. 2023A Comprehensive Overview of Large Language Modelscited by 639
  12. 2023Instruction Tuning for Large Language Models: A Surveycited by 617
38 more
  1. 2024ORPO: Monolithic Preference Optimization without Reference Modelcited by 556
  2. 2023MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuningcited by 483
  3. 2023Model Tells You What to Discard: Adaptive KV Cache Compression for LLMscited by 466
  4. 2024Safety Alignment Should Be Made More Than Just a Few Tokens Deepcited by 450
  5. 2023Catastrophic Jailbreak of Open-source LLMs via Exploiting Generationcited by 439
  6. 2023How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resourcescited by 405
  7. 2023GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Ciphercited by 401
  8. 2023Scaling Data-Constrained Language Modelscited by 391
  9. 2023ToolAlpaca: Generalized Tool Learning for Language Models with 3000 Simulated Casescited by 383
  10. 2024Hallucination of Multimodal Large Language Models: A Surveycited by 380
  11. 2023Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructionscited by 377
  12. 2024Mini-Gemini: Mining the Potential of Multi-modality Vision Language Modelscited by 361
  13. 2024ReST-MCTS*: LLM Self-Training via Process Reward Guided Tree Searchcited by 338
  14. 2023Explainability for Large Language Models: A Surveycited by 331
  15. 2024What matters when building vision-language models?cited by 321
  16. 2023From Quantity to Quality: Boosting LLM Performance with Self-Guided Data Selection for Instruction Tuningcited by 301
  17. 2024Aya Model: An Instruction Finetuned Open-Access Multilingual Language Modelcited by 292
  18. 2023Alphazero-like Tree-Search can Guide Large Language Model Decoding and Trainingcited by 291
  19. 2024OLMoE: Open Mixture-of-Experts Language Modelscited by 291
  20. 2023Evaluating Large Language Models at Evaluating Instruction Followingcited by 290
  21. 2023Shadow Alignment: The Ease of Subverting Safely-Aligned Language Modelscited by 284
  22. 2023Preference Ranking Optimization for Human Alignmentcited by 283
  23. 2024A Survey on Knowledge Distillation of Large Language Modelscited by 276
  24. 2024Knowledge Conflicts for LLMs: A Surveycited by 274
  25. 2023RAGTruth: A Hallucination Corpus for Developing Trustworthy Retrieval-Augmented Language Modelscited by 272
  26. 2023Challenges and Applications of Large Language Modelscited by 251
  27. 2023The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learningcited by 251
  28. 2023OpenChat: Advancing Open-source Language Models with Mixed-Quality Datacited by 250
  29. 2024Playground v2.5: Three Insights towards Enhancing Aesthetic Quality in Text-to-Image Generationcited by 249
  30. 2023FacTool: Factuality Detection in Generative AI -- A Tool Augmented Framework for Multi-Task and Multi-Domain Scenarioscited by 247
  31. 2023Multimodal Foundation Models: From Specialists to General-Purpose Assistantscited by 247
  32. 2023FireAct: Toward Language Agent Fine-tuningcited by 247
  33. 2023Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Modelscited by 237
  34. 2024SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decodingcited by 233
  35. 2024Generative Representational Instruction Tuningcited by 227
  36. 2024Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?cited by 227
  37. 2023Beyond Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimizationcited by 223
  38. 2023How Abilities in Large Language Models are Affected by Supervised Fine-tuning Data Compositioncited by 222

50 most cited of 1,080 All 1,080 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Aligning Large Language Models via Fine-grained Supervision
  2. 2025StatLLaMA: Multi-Stage training for domain-optimized statistical large language models
  3. 2025REFINE-AF: A Task-Agnostic Framework to Align Language Models via Self-Generated Instructions using Reinforcement Learning from Automated Feedback
  4. 2025Aligning Instruction Tuning with Pre-training
  5. 2024Accelerating Large Language Model Pretraining via LFR Pedagogy: Learn, Focus, and Review
  6. 2023An Emulator for Fine-Tuning Large Language Models using Small Language Models
  7. 2024InstructionCP: A fast approach to transfer Large Language Models into target language
  8. 2024Reinforcement Learning for Aligning Large Language Models Agents with Interactive Environments: Quantifying and Mitigating Prompt Overfitting