weili-0234/qatfactory-gptq-calibration
QATFactory GPTQ calibration set (calib_512) This calibration dataset is a chosen subset of OpenPerfectBlend. Concretely: 512 rows sampled with random.Random(42).sample(range(n_rows), 512) (then sorted) from openperfectblend_100k_Qwen3.5-9B_think_train.jsonl — a 100K-example OpenPerfectBlend prompt set whose responses (including <think> reasoning traces) were generated by Qwen/Qwen3.5-9B. Each row is the conversation rendered with the model's stock chat template… See the full description on the dataset page: https://huggingface.co/datasets/weili-0234/qatfactory-gptq-calibration.
031
Upload README.md with huggingface_hub
Upload calib_512.meta.json with huggingface_hub
Upload calib_512.jsonl with huggingface_hub
initial commit
