Complexity-aware fine-tuning (2506.21220v1)

Abstract: General-purpose LLMs are frequently fine-tuned through supervised fine-tuning (SFT) to enhance performance in specific domains. Better results can be achieved by distilling the chain-of-thought of a larger model at the cost of numerous expensive calls and a much greater amount of data. We propose a novel blueprint for efficient fine-tuning that uses reasoning only for complex data identified by entropy. Specifically, across two small open models ($\approx 3B$) we split the training data into complexity categories by a single token answer entropy (ROC AUC $0.73$), fine-tune LLMs via SFT and distillation, and show that our pipeline significantly outperforms the standard SFT approach ($0.55$ vs $0.43$ average accuracy) and provides comparable with distillation performance while using $62\%$ less data ($0.55$ average accuracy for both). We publish our code and data to facilitate further research in this direction.