Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2510.05871 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908579541286912 |
|---|---|
| author | Stoisser, Josefa Lia Phillips, Lawrence Misra, Aditya Lamb, Tom A. Torr, Philip Martell, Marc Boubnovski Fauqueur, Julien Märtens, Kaspar |
| author_facet | Stoisser, Josefa Lia Phillips, Lawrence Misra, Aditya Lamb, Tom A. Torr, Philip Martell, Marc Boubnovski Fauqueur, Julien Märtens, Kaspar |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_05871 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Towards Label-Free Biological Reasoning Synthetic Dataset Creation via Uncertainty Filtering Stoisser, Josefa Lia Phillips, Lawrence Misra, Aditya Lamb, Tom A. Torr, Philip Martell, Marc Boubnovski Fauqueur, Julien Märtens, Kaspar Artificial Intelligence Machine Learning 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. |
| title | Towards Label-Free Biological Reasoning Synthetic Dataset Creation via Uncertainty Filtering |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2510.05871 |