Search papers, labs, and topics across Lattice.
The paper introduces Logics-STEM, a reasoning model fine-tuned on a 10M-scale Logics-STEM-SFT dataset, designed for STEM reasoning tasks. The model achieves a 4.68% average improvement over other 8B models on STEM benchmarks, attributed to a data-algorithm co-design approach involving meticulous data curation and failure-driven post-training. This post-training framework uses targeted knowledge retrieval and data synthesis to address model failure regions during supervised fine-tuning.
Failure-driven post-training, combined with a meticulously curated 10M token STEM dataset, unlocks a 4.68% performance boost in LLM reasoning, proving that strategic data synthesis around model weaknesses is a powerful path to improvement.
We present Logics-STEM, a state-of-the-art reasoning model fine-tuned on Logics-STEM-SFT-Dataset, a high-quality and diverse dataset at 10M scale that represents one of the largest-scale open-source long chain-of-thought corpora. Logics-STEM targets reasoning tasks in the domains of Science, Technology, Engineering, and Mathematics (STEM), and exhibits exceptional performance on STEM-related benchmarks with an average improvement of 4.68% over the next-best model at 8B scale. We attribute the gains to our data-algorithm co-design engine, where they are jointly optimized to fit a gold-standard distribution behind reasoning. Data-wise, the Logics-STEM-SFT-Dataset is constructed from a meticulously designed data curation engine with 5 stages to ensure the quality, diversity, and scalability, including annotation, deduplication, decontamination, distillation, and stratified sampling. Algorithm-wise, our failure-driven post-training framework leverages targeted knowledge retrieval and data synthesis around model failure regions in the Supervised Fine-tuning (SFT) stage to effectively guide the second-stage SFT or the reinforcement learning (RL) for better fitting the target distribution. The superior empirical performance of Logics-STEM reveals the vast potential of combining large-scale open-source data with carefully designed synthetic data, underscoring the critical role of data-algorithm co-design in enhancing reasoning capabilities through post-training. We make both the Logics-STEM models (8B and 32B) and the Logics-STEM-SFT-Dataset (10M and downsampled 2.2M versions) publicly available to support future research in the open-source community.
DAMO