API-guided Dataset Synthesis to Finetune Large Code Models

• URL: http://arxiv.org/abs/2408.08343v2
• Date: Thu, 22 Aug 2024 11:29:51 GMT
• Title: API-guided Dataset Synthesis to Finetune Large Code Models
• Authors: Zongjie Li, Daoyuan Wu, Shuai Wang, Zhendong Su,
• Abstract summary: Large code models (LCMs) are pre-trained on vast code corpora. Supervised fine-tuning (SFT) plays a vital role in aligning these models with specific requirements. High-quality SFT datasets pose a significant challenge due to the uneven quality of datasets and the scarcity of domain-specific datasets. We propose DataScope, an API-guided dataset synthesis framework.
• Score: 13.835835256858653
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large code models (LCMs), pre-trained on vast code corpora, have demonstrated remarkable performance across a wide array of code-related tasks. Supervised fine-tuning (SFT) plays a vital role in aligning these models with specific requirements and enhancing their performance in particular domains. However, synthesizing high-quality SFT datasets poses a significant challenge due to the uneven quality of datasets and the scarcity of domain-specific datasets. Inspired by APIs as high-level abstractions of code that encapsulate rich semantic information in a concise structure, we propose DataScope, an API-guided dataset synthesis framework designed to enhance the SFT process for LCMs in both general and domain-specific scenarios. DataScope comprises two main components: Dsel and Dgen. On one hand, Dsel employs API coverage as a core metric, enabling efficient dataset synthesis in general scenarios by selecting subsets of existing (uneven-quality) datasets with higher API coverage. On the other hand, Dgen recasts domain dataset synthesis as a process of using API-specified high-level functionality and deliberately-constituted code skeletons to synthesize concrete code. Extensive experiments demonstrate DataScope's effectiveness, with models fine-tuned on its synthesized datasets outperforming those tuned on unoptimized datasets five times larger. Furthermore, a series of analyses on model internals, relevant hyperparameters, and case studies provide additional evidence for the efficacy of our proposed methods. These findings underscore the significance of dataset quality in SFT and advance the field of LCMs by providing an efficient, cost-effective framework for constructing high-quality datasets. This contribution enhances performance across both general and domain-specific scenarios, paving the way for more powerful and tailored LCMs.

Related papers

• AQuilt: Weaving Logic and Self-Inspection into Low-Cost, High-Relevance Data Synthesis for Specialist LLMs [24.511434195711267]
  AQuilt is a framework for constructing instruction-tuning data for any specialized domains from corresponding unlabeled data.<n>We construct a dataset of 703k examples to train a powerful data synthesis model.
  arXiv  Detail & Related papers  (2025-07-24T17:03:27Z)
• RouteNator: A Router-Based Multi-Modal Architecture for Generating Synthetic Training Data for Function Calling LLMs [3.41612427812159]
  In digital content creation tools, users express their needs through natural language queries that must be mapped to API calls.<n>Existing approaches to synthetic data generation fail to replicate real-world data distributions.<n>We present a novel router-based architecture that generates high-quality synthetic training data.
  arXiv  Detail & Related papers  (2025-05-15T16:53:45Z)
• Scaling Laws of Synthetic Data for Language Models [132.67350443447611]
  We introduce SynthLLM, a scalable framework that transforms pre-training corpora into diverse, high-quality synthetic datasets. Our approach achieves this by automatically extracting and recombining high-level concepts across multiple documents using a graph algorithm.
  arXiv  Detail & Related papers  (2025-03-25T11:07:12Z)
• Unsupervised Topic Models are Data Mixers for Pre-training Language Models [6.77198566340415]
  We propose a topic-based data mixing strategy for large language models (LLMs) DataWeave employs a multi-stage clustering process to group semantically similar documents. We confirm that the topics Science and Relationships are particularly effective, yielding the most substantial performance improvements.
  arXiv  Detail & Related papers  (2025-02-24T03:25:56Z)
• mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data [71.352883755806]
  Multimodal embedding models have gained significant attention for their ability to map data from different modalities, such as text and images, into a unified representation space. However, the limited labeled multimodal data often hinders embedding performance. Recent approaches have leveraged data synthesis to address this problem, yet the quality of synthetic data remains a critical bottleneck.
  arXiv  Detail & Related papers  (2025-02-12T15:03:33Z)
• TARGA: Targeted Synthetic Data Generation for Practical Reasoning over Structured Data [9.390415313514762]
  TARGA is a framework that generates high-relevance synthetic data without manual annotation. It substantially outperforms existing non-fine-tuned methods that utilize close-sourced model. It exhibits superior sample efficiency, robustness, and generalization capabilities under non-I.I.D. settings.
  arXiv  Detail & Related papers  (2024-12-27T09:16:39Z)
• Data-Juicer 2.0: Cloud-Scale Adaptive Data Processing for and with Foundation Models [64.28420991770382]
  Data-Juicer 2.0 is a data processing system backed by data processing operators spanning text, image, video, and audio modalities.<n>It supports more critical tasks including data analysis, annotation, and foundation model post-training.<n>It has been widely adopted in diverse research fields and real-world products such as Alibaba Cloud PAI.
  arXiv  Detail & Related papers  (2024-12-23T08:29:57Z)
• 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)
• Generating Realistic Tabular Data with Large Language Models [49.03536886067729]
  Large language models (LLM) have been used for diverse tasks, but do not capture the correct correlation between the features and the target variable. We propose a LLM-based method with three important improvements to correctly capture the ground-truth feature-class correlation in the real data. Our experiments show that our method significantly outperforms 10 SOTA baselines on 20 datasets in downstream tasks.
  arXiv  Detail & Related papers  (2024-10-29T04:14:32Z)
• Efficacy of Synthetic Data as a Benchmark [3.2968976262860408]
  We investigate the effectiveness of generating synthetic data through large language models (LLMs) Our experiments show that while synthetic data can effectively capture performance of various methods for simpler tasks, it falls short for more complex tasks like named entity recognition. We propose a new metric called the bias factor, which evaluates the biases introduced when the same LLM is used to both generate benchmarking data and to perform the tasks.
  arXiv  Detail & Related papers  (2024-09-18T13:20:23Z)
• Unveiling the Impact of Coding Data Instruction Fine-Tuning on Large Language Models Reasoning [64.5243480989869]
  Instruction Fine-Tuning (IFT) significantly enhances the zero-shot capabilities of pretrained Large Language Models (LLMs) This paper investigates how coding data impact LLMs' reasoning capacities during the IFT stage.
  arXiv  Detail & Related papers  (2024-05-30T23:20:25Z)
• Dial-insight: Fine-tuning Large Language Models with High-Quality Domain-Specific Data Preventing Capability Collapse [4.98050508891467]
  We propose a two-stage approach for the construction of production prompts designed to yield high-quality data. This method involves the generation of a diverse array of prompts that encompass a broad spectrum of tasks and exhibit a rich variety of expressions. We introduce a cost-effective, multi-dimensional quality assessment framework to ensure the integrity of the generated labeling data.
  arXiv  Detail & Related papers  (2024-03-14T08:27:32Z)
• Distribution-Aware Data Expansion with Diffusion Models [55.979857976023695]
  We propose DistDiff, a training-free data expansion framework based on the distribution-aware diffusion model. DistDiff consistently enhances accuracy across a diverse range of datasets compared to models trained solely on original data.
  arXiv  Detail & Related papers  (2024-03-11T14:07:53Z)
• LESS: Selecting Influential Data for Targeted Instruction Tuning [64.78894228923619]
  We propose LESS, an efficient algorithm to estimate data influences and perform Low-rank gradiEnt Similarity Search for instruction data selection. We show that training on a LESS-selected 5% of the data can often outperform training on the full dataset across diverse downstream tasks. Our method goes beyond surface form cues to identify data that the necessary reasoning skills for the intended downstream application.
  arXiv  Detail & Related papers  (2024-02-06T19:18:04Z)
• TRoVE: Transforming Road Scene Datasets into Photorealistic Virtual Environments [84.6017003787244]
  This work proposes a synthetic data generation pipeline to address the difficulties and domain-gaps present in simulated datasets. We show that using annotations and visual cues from existing datasets, we can facilitate automated multi-modal data generation.
  arXiv  Detail & Related papers  (2022-08-16T20:46:08Z)
• SYNC: A Copula based Framework for Generating Synthetic Data from Aggregated Sources [8.350531869939351]
  We study synthetic data generation task called downscaling. We propose a multi-stage framework called SYNC (Synthetic Data Generation via Gaussian Copula) We make four key contributions in this work.
  arXiv  Detail & Related papers  (2020-09-20T16:36:25Z)

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.