Podtech/Swallow-Nemotron-Post-Training-Dataset-v1-ja-cpt

Swallow-Nemotron-Post-Training-Dataset-v1-ja-cpt Dataset Overview This dataset is a reformatted subset of the tokyotech-llm/Swallow-Nemotron-Post-Training-Dataset-v1 dataset, specifically derived from the v1-Ja-202601 subset. It was created to facilitate Continuous Pre-Training (CPT) by extracting only the text_gpt_oss field from the original data. Dataset Statistics & Token Counts The token counts for each category were calculated using the… See the full description on the dataset page: https://huggingface.co/datasets/Podtech/Swallow-Nemotron-Post-Training-Dataset-v1-ja-cpt.