mlsanityData to 7 Oct 2026

Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning

arXiv 2311.08182 ↗First posted 14 Nov 2023cs.CL cs.LGcited by 52

On the map: Synthetic data and self-improvement for LLMs → Code and instruction data synthesis

Abstract (arXiv)

Enhancing the instruction-following ability of Large Language Models (LLMs) primarily demands substantial instruction-tuning datasets. However, the sheer volume of these imposes a considerable computational burden and annotation cost. To investigate a label-efficient instruction tuning method that allows the model itself to actively sample subsets that are equally or even more effective, we introduce a self-evolving mechanism DiverseEvol. In this process, a model iteratively augments its training subset to refine its own performance, without requiring any intervention from humans or more advanced LLMs. The key to our data sampling technique lies in the enhancement of diversity in the chosen subsets, as the model selects new data points most distinct from any existing ones according to its current embedding space. Extensive experiments across three datasets and benchmarks demonstrate the effectiveness of DiverseEvol. Our models, trained on less than 8% of the original dataset, maintain or improve performance compared with finetuning on full data. We also provide empirical evidence to analyze the importance of diversity in instruction data and the iterative scheme as opposed to one-time sampling. Our code is publicly available at https://github.com/OFA-Sys/DiverseEvol.git.

Key facts. Cited by 52 papers in the map (data through 7 Oct 2026). Placed in Code and instruction data synthesis, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 8 of its 24 references that are in the map. First posted on arXiv 14 Nov 2023.

Builds on

8 of 24 references.

Closest ancestors · 1

Same direction, newest first

  1. 2023Becoming self-instruct: introducing early stopping criteria for minimal instruct tuning5 Jul

Other references · 7

  1. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  2. 2023WizardLM: Empowering large pre-trained language models to follow complex instructionscited by 1,116
  3. 2023AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedbackcited by 738
  4. 2023OpenAssistant Conversations -- Democratizing Large Language Model Alignmentcited by 712
  5. 2022The Vendi Score: A Diversity Evaluation Metric for Machine Learningcited by 299
  6. 2023ExpertPrompting: Instructing Large Language Models to be Distinguished Expertscited by 140
  7. 2021Batch Active Learning at Scalecited by 127

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 24; 2025: 19; 2026: 8.

52 papers.

Sort:

By area:

  1. 2025Large Language Model Agent: A Survey on Methodology, Applications and Challengescited by 194
  2. 2024A Survey on Self-Evolution of Large Language Modelscited by 86
  3. 2024A Survey on Data Selection for LLM Instruction Tuningcited by 76
  4. 2025From AI for Science to Agentic Science: A Survey on Autonomous Scientific Discoverycited by 74
  5. 2024Self-Exploring Language Models: Active Preference Elicitation for Online Alignmentcited by 63
  6. 2025Ming-Omni: A Unified Multimodal Model for Perception and Generationcited by 62
  7. 2024KnowAgent: Knowledge-Augmented Planning for LLM-Based Agentscited by 61
  8. 2024SmallToLarge (S2L): Scalable Data Selection for Fine-tuning Large Language Models by Summarizing Training Trajectories of Small Modelscited by 43
  9. 2025Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoningcited by 38
  10. 2025Principled Data Selection for Alignment: The Hidden Risks of Difficult Examplescited by 37
  11. 2024SelectIT: Selective Instruction Tuning for LLMs via Uncertainty-Aware Self-Reflectioncited by 34
  12. 2026LongCat-Flash-Thinking-2601 Technical Reportcited by 32
38 more
  1. 2024Causal Prompting: Debiasing Large Language Model Prompting based on Front-Door Adjustmentcited by 29
  2. 2024Unleashing the Power of Data Tsunami: A Comprehensive Survey on Data Assessment and Selection for Instruction Tuning of Language Modelscited by 29
  3. 2024Online Merging Optimizers for Boosting Rewards and Mitigating Tax in Alignmentcited by 28
  4. 2024Concept-skill Transferability-based Data Selection for Large Vision-Language Modelscited by 24
  5. 2025Measuring Data Diversity for Instruction Tuning: A Systematic Analysis and A Reliable Metriccited by 24
  6. 2024Rethinking Data Selection at Scale: Random Selection is Almost All You Needcited by 22
  7. 2024SHED: Shapley-Based Automated Dataset Refinement for Instruction Fine-Tuningcited by 19
  8. 2023Data Management For Training Large Language Models: A Surveycited by 17
  9. 2025Toward Effective Tool-Integrated Reasoning via Self-Evolved Preference Learningcited by 17
  10. 2024SMART: Submodular Data Mixture Strategy for Instruction Tuningcited by 16
  11. 2024Quality-Weighted Vendi Scores And Their Application To Diverse Experimental Designcited by 15
  12. 2024Exploring the Mystery of Influential Data for Mathematical Reasoningcited by 13
  13. 2025LEAD: Iterative Data Selection for Efficient LLM Instruction Tuningcited by 13
  14. 2025MIG: Automatic Data Selection for Instruction Tuning by Maximizing Information Gain in Semantic Spacecited by 12
  15. 2024LLMs-as-Instructors: Learning from Errors Toward Automating Model Improvementcited by 11
  16. 2024Mastering Collaborative Multi-modal Data Selection: A Focus on Informativeness, Uniqueness, and Representativenesscited by 10
  17. 2024Federated Data-Efficient Instruction Tuning for Large Language Modelscited by 9
  18. 2025Difficulty-Based Preference Data Selection by DPO Implicit Reward Gapcited by 7
  19. 2025Importance-Aware Data Selection for Efficient LLM Instruction Tuningcited by 7
  20. 2024Baichuan Alignment Technical Reportcited by 6
  21. 2026Self-Improvement of Large Language Models: A Technical Overview and Future Outlookcited by 5
  22. 2024Instruction Embedding: Latent Representations of Instructions Towards Task Identificationcited by 4
  23. 2025Forgetting: A New Mechanism Towards Better Large Language Model Fine-tuningcited by 4
  24. 2025Towards Understanding Valuable Preference Data for Large Language Model Alignmentcited by 4
  25. 2025MDIT: A Model-free Data Interpolation Method for Diverse Instruction Tuningcited by 3
  26. 2025ReSURE: Regularizing Supervision Unreliability for Multi-turn Dialogue Fine-tuningcited by 3
  27. 2024Leveraging Open Knowledge for Advancing Task Expertise in Large Language Modelscited by 2
  28. 2025Teaching According to Talents! Instruction Tuning LLMs with Competence-Aware Curriculum Learningcited by 2
  29. 2024Data Quality Control in Federated Instruction-tuning of Large Language Modelscited by 1
  30. 2025InfiAlign: A Scalable and Sample-Efficient Framework for Aligning LLMs to Enhance Reasoning Capabilitiescited by 1
  31. 2026SynAE: A Framework for Measuring the Quality of Synthetic Data for Tool-Calling Agent Evaluationscited by 1
  32. 2026Mechanistic Circuit Identification for Controllable Data Generationcited by 1
  33. 2024EasyInstruct: An Easy-to-use Instruction Processing Framework for Large Language Modelscited by 0
  34. 2024A Framework for Fine-Tuning LLMs using Heterogeneous Feedbackcited by 0
  35. 2025InsBank: Evolving Instruction Subset for Ongoing Alignmentcited by 0
  36. 2025TACOS: Open Tagging and Comparative Scoring for Instruction Fine-Tuning Data Selectioncited by 0
  37. 2026EvoSelect: Data-Efficient LLM Evolution for Targeted Task Adaptationcited by 0
  38. 2026GRACE: Gradient-aligned Reasoning Data Curation for Efficient Post-trainingcited by 0

50 most cited of 52 All 52 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Aligning Instruction Tuning with Pre-training
  2. 2024Optimizing Instruction Synthesis: Effective Exploration of Evolutionary Space with Tree Search
  3. 2025CrowdSelect: Synthetic Instruction Data Selection with Multi-LLM Wisdom
  4. 2025Towards Alignment-Centric Paradigm: A Survey of Instruction Tuning in Large Language Models
  5. 2024Diversify and Conquer: Diversity-Centric Data Selection with Iterative Refinement
  6. 2024LESS: Selecting Influential Data for Targeted Instruction Tuning
  7. 2025MAIN: Mutual Alignment Is Necessary for instruction tuning
  8. 2024ROSE: A Reward-Oriented Data Selection Framework for LLM Task-Specific Instruction Tuning