This repository is a convenience collection of the NVIDIA Nemotron Pretraining Specialized releases. It preserves each original subset as a separate Hugging Face configuration, so consumers can select a single domain-focused subset without visiting multiple source repositories. Every configuration has a train split. Configuration names begin with the source release (v1, v1.1, or v1.2) to keep similarly named subsets unambiguous. Replace the configuration name with any name listed in the repository’s configuration selector. Schemas can differ by source subset; inspect dataset.features after loading rather than assuming a common schema. The records are redistributed from the source releases…
Publicly accessible
cc-by-4.0
CrowdMind/SFT-Combinded-USE is a convenience collection of supervised fine-tuning, tool-use, coding, and agent-trace datasets. Source subsets remain separate configurations: no rows are merged, transformed, or forced into a shared schema. Configuration names are short and readable. other, tooluse, and traces identify the source collection; the remaining terms identify the upstream dataset and split. Inspect each configuration's features before use. Their schemas intentionally differ. This repository does not claim authorship of the upstream data or create new annotations. Licenses, attribution requirements, access restrictions, and redistribution permissions vary by source. Users must…
Publicly accessible
