EDCO: Dynamic Curriculum Orchestration for Domain-specific Large Language Model Fine-tuning

Domain-specific large language models (LLMs), typically developed by fine-tuning a pre-trained general-purpose LLM on specialized datasets, represent a significant advancement in applied AI. A common strategy in LLM fine-tuning is curriculum learning, which pre-orders training samples based on metrics like difficulty to improve learning efficiency compared to a random sampling strategy. However, most existing methods for LLM fine-tuning rely on a static curriculum, designed prior to training, which lacks adaptability to the model's evolving needs during fine-tuning. To address this, we propose EDCO, a novel framework based on two key concepts: \textit{inference entropy} and \textit{dynamic curriculum orchestration}. Inspired by recent findings that maintaining high answer entropy benefits long-term reasoning gains, EDCO prioritizes samples with high inference entropy in a continuously adapted curriculum. EDCO integrates three core components: an efficient entropy estimator that uses prefix tokens to approximate full-sequence entropy, an entropy-based curriculum generator that selects data points with the highest inference entropy, and an LLM trainer that optimizes the model on the selected curriculum. Comprehensive experiments in communication, medicine and law domains, EDCO outperforms traditional curriculum strategies for fine-tuning Qwen3-4B and Llama3.2-3B models under supervised and reinforcement learning settings. Furthermore, the proposed efficient entropy estimation reduces computational time by 83.5\% while maintaining high accuracy. The code is available at https://github.com/GTS-AIData/EDCO.

Large language models often need extra training before they can work well in specialized areas such as medicine, law, and communications. However, high-quality training data in these areas is usually limited, so it matters which examples the model learns from and when it sees them. This paper introduces EDCO, a method that keeps updating the training examples based on what the model currently finds uncertain. Instead of following a fixed training order, EDCO repeatedly looks for examples that still challenge the model and uses them to guide the next stage of training. This helps the model spend more effort on examples that are likely to teach it something new. We also make this process efficient by estimating uncertainty from only the beginning of the model’s answer, rather than from a full response. Experiments across communications, medicine, law, math reasoning, and summarization show that EDCO improves performance over random training and several other ways of choosing training data, while adding only modest extra computation.