{"ID":2859323,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.05871","arxiv_id":"2510.05871","title":"Towards Label-Free Biological Reasoning Synthetic Dataset Creation via Uncertainty Filtering","abstract":"Synthetic chain-of-thought (CoT) traces are widely used to train large reasoning models (LRMs), improving generalization by providing step-level supervision. Yet most approaches require ground-truth labels to seed or filter these traces - an expensive bottleneck in domains like biology where wet-lab data are scarce. We propose a label-free alternative: uncertainty-based filtering, which uses a model's own confidence - quantified through established uncertainty metrics like self-consistency and predictive perplexity - as a substitute for external labels. We sample multiple reasoning traces and retain only low-uncertainty subsets. Applied to biological perturbation prediction, a domain where wet-lab labels are especially costly, we show that the filtered subset has higher accuracy, and that supervised fine-tuning (SFT) on uncertainty-filtered data outperforms unfiltered synthetic data, narrows the gap to ground-truth training, and surpasses strong LRM baselines. Ablations show that per-class filtering corrects for class-specific uncertainty scales and that hybrid uncertainty metrics yield higher-quality datasets. Our results suggest that model-internal confidence is a powerful signal for efficient reasoning dataset creation, enabling LRMs in domains where supervision is expensive.","short_abstract":"Synthetic chain-of-thought (CoT) traces are widely used to train large reasoning models (LRMs), improving generalization by providing step-level supervision. Yet most approaches require ground-truth labels to seed or filter these traces - an expensive bottleneck in domains like biology where wet-lab data are scarce. We...","url_abs":"https://arxiv.org/abs/2510.05871","url_pdf":"https://arxiv.org/pdf/2510.05871v1","authors":"[\"Josefa Lia Stoisser\",\"Lawrence Phillips\",\"Aditya Misra\",\"Tom A. Lamb\",\"Philip Torr\",\"Marc Boubnovski Martell\",\"Julien Fauqueur\",\"Kaspar Märtens\"]","published":"2025-10-07T12:40:37Z","proceeding":"cs.AI","tasks":"[\"cs.AI\",\"cs.LG\"]","methods":"[]","has_code":false}