2026-10-12T03:09:50Z https://www.scidb.cn/oai

10.57760/sciencedb.j00133.00710 2026-09-20T09:58:14Z

2026-09-20 CAMEL-7B Multi-Agent Instruction Tuning Dataset doi:10.57760/sciencedb.j00133.00710 en The CAMEL-7B Multi-Agent Instruction Fine-tuning Dataset aims to address the high cost of acquiring high-quality instruction fine-tuning data and the degradation issues associated with automatic data generation. Based on an improved CAMEL framework, the data was generated by introducing a behavioral constraint mechanism, a predefined library of specialized roles (e.g., Product Managers, Engineers), and structured interaction protocols to produce task-oriented multi-turn collaborative dialogues. Grounded on 1,000 groups of high-quality multi-turn dialogues, the dataset was refined through cleaning and structured extraction to form 2,315 standardized instruction-response pairs. The data covers four main domains: code generation (45%), mathematical reasoning (30%), scientific research tasks (15%), and other comprehensive fields (10%). Experiments demonstrate that applying supervised fine-tuning (SFT) with this dataset significantly improves the performance of 1.3B to 7B scale open-source models on benchmarks such as HumanEval, exhibiting strong cross-model generalization capabilities. Large Language Models; Instruction Fine-tuning; Multi-agent Collaboration Jason PUBLIC https://creativecommons.org/publicdomain/zero/1.0/ dataset Science Data Bank