A curated, unified SFT dataset distilled from frontier model traces across multiple architectures and reasoning styles. Named after the accretion disk — the region where matter spirals inward, accelerated and organized by gravity into a coherent structure. This dataset does the same for language model training: it takes high-quality traces from frontier models and accelerates them into a focused, well-organized training resource.… See the full description on the dataset page:
https://huggingface.co/datasets/predictive-singularity/Accretion.
