DataEvolver: Automatic Data Preparation for Large Language Models through Multi-Level Self-Evolving

• URL: http://arxiv.org/abs/2606.07001v2
• Date: Wed, 10 Jun 2026 07:54:44 GMT
• Title: DataEvolver: Automatic Data Preparation for Large Language Models through Multi-Level Self-Evolving
• Abstract summary: High-quality training data is essential to large language models (LLMs)<n>Existing automatic data preparation methods rely on predefined pipelines or customized human instructions.<n>We introduce DataEvolver, the first self-evolving data preparation system that automatically constructs pipelines to transform raw data into high-quality data.
• Score: 55.85201665674226
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: High-quality training data is essential to large language models (LLMs) and typically requires extensive and costly manual curation. Existing automatic data preparation methods rely on predefined pipelines or customized human instructions, which limits their adaptability to diverse data distributions and lacks principled guidance from high-quality examples. In this paper, we introduce DataEvolver, the first self-evolving data preparation system that automatically constructs pipelines to transform raw data into high-quality data. DataEvolver employs a multi-level mechanism to ensure both pipeline executability and effectiveness. At the operator level, it incrementally expands the operator set to construct a logical plan while resolving dependency conflicts. At the pipeline level, it instantiates logical plans into executable code and iteratively refines pipeline orchestration through a feedback loop that reduces the distribution gap between prepared data and high-quality examples. Experiments on seven benchmarks show that DataEvolver substantially improves data quality and achieves an average 10\% gain in downstream LLM performance compared with training on original data, highlighting new opportunities for the iterative co-evolution of LLMs and data.

Related papers

• DataFoundry: Evolving Data Preparators via Recursive Self-Improvement [34.053745857869636]
  textscDataFoundry represents a data preparator as an evolvable runtime specification and instantiates its evolution with a textsc-as-Modules architecture.<n>We evaluate textscDataFoundry on DataPrep-Bench across mathematics, finance, law, and medicine.
  arXiv  Detail & Related papers  (2026-08-30T18:53:24Z)
• DataClaw0: Agentic Tailoring Multimodal Data from Raw Streams [48.434767119224375]
  We propose a paradigm shift towards Agentic Data Tailoring.<n>To overcome the data scarcity bottleneck in training such high-order capabilities, we design a two-stage pipeline.<n>$textDataClaw_0$-9B model synergizes Supervised Fine-Tuning with Group Relative Policy Optimization.
  arXiv  Detail & Related papers  (2026-06-19T11:31:43Z)
• Towards Next-Generation LLM Training: From the Data-Centric Perspective [29.115711465172122]
  Large language models (LLMs) have demonstrated remarkable performance across a wide range of tasks and domains, with data playing a central role in enabling these advances.<n>Despite this success, the preparation and effective utilization of the massive datasets required for LLM training remain major bottlenecks.<n>We propose building a robust, agent-based automatic data preparation system that supports automated workflow construction and scalable data management.<n>We argue for a unified data-model interaction training system in which data is dynamically selected, mixed, and reweighted throughout the training process, enabling more efficient, adaptive, and performance-aware data utilization.
  arXiv  Detail & Related papers  (2026-03-16T01:40:09Z)
• Data Science and Technology Towards AGI Part I: Tiered Data Management [53.64581824953229]
  We argue that the development of artificial intelligence is entering a new phase of data-model co-evolution.<n>We introduce an L0-L4 tiered data management framework, ranging from raw uncurated resources to organized and verifiable knowledge.<n>We validate the effectiveness of the proposed framework through empirical studies.
  arXiv  Detail & Related papers  (2026-02-09T18:47:51Z)
• LLM-AutoDP: Automatic Data Processing via LLM Agents for Model Fine-tuning [12.792070502265616]
  Large Language Models (LLMs) can be fine-tuned on domain-specific data to enhance their performance in specialized fields.<n>Such data often contains numerous low-quality samples, necessitating effective data processing (DP)
  arXiv  Detail & Related papers  (2026-01-28T08:37:34Z)
• Closing the Data Loop: Using OpenDataArena to Engineer Superior Training Datasets [46.480867560675584]
  We propose a paradigm shift from ad-hoc curation to a closed-loop dataset engineering framework using OpenDataArena (ODA)<n>We instantiate this methodology through two new datasets: textbfODA-Math460-k, a specialized mathematics reasoning dataset that utilizes a novel two-stage difficulty-aware pipeline to achieve State-of-the-Art (SOTA) results on benchmarks such as AIME and HMMT, and textbfODA-Mixture (100k & 500k), a series of multi-domain instruction datasets built via an Anchor-and-
  arXiv  Detail & Related papers  (2025-12-30T17:46:38Z)
• DataFlow: An LLM-Driven Framework for Unified Data Preparation and Workflow Automation in the Era of Data-Centric AI [42.191938707504406]
  DataFlow is a unified and LLM-driven data preparation framework.<n>System-level abstractions enable modular, reusable, and composable data transformations.<n>DataFlow consistently improves downstream Large Language Models performance.
  arXiv  Detail & Related papers  (2025-12-18T15:46:15Z)
• Aligning Instruction Tuning with Pre-training [61.50161961371844]
  We propose Aligning Instruction Tuning with Pre-training (AITP) to align instruction tuning with pre-training distributions.<n>We show consistent performance improvements with AITP on three fully open large language models (LLMs) across eight benchmarks.
  arXiv  Detail & Related papers  (2025-01-16T08:27:40Z)
• Star-Agents: Automatic Data Optimization with LLM Agents for Instruction Tuning [71.2981957820888]
  We propose a novel Star-Agents framework, which automates the enhancement of data quality across datasets. The framework initially generates diverse instruction data with multiple LLM agents through a bespoke sampling method. The generated data undergo a rigorous evaluation using a dual-model method that assesses both difficulty and quality.
  arXiv  Detail & Related papers  (2024-11-21T02:30:53Z)
• RECOST: External Knowledge Guided Data-efficient Instruction Tuning [25.985023475991625]
  We argue that most current data-efficient instruction-tuning methods are highly dependent on the quality of the original instruction-tuning dataset. We propose a framework dubbed as textbfRECOST, which integrates external-knowledge-base re-ranking and diversity-consistent sampling into a single pipeline.
  arXiv  Detail & Related papers  (2024-02-27T09:47:36Z)
• STAR: Boosting Low-Resource Information Extraction by Structure-to-Text Data Generation with Large Language Models [56.27786433792638]
  STAR is a data generation method that leverages Large Language Models (LLMs) to synthesize data instances. We design fine-grained step-by-step instructions to obtain the initial data instances. Our experiments show that the data generated by STAR significantly improve the performance of low-resource event extraction and relation extraction tasks.
  arXiv  Detail & Related papers  (2023-05-24T12:15:19Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.