CoolFace
Datasetpublicgated

slippedJim/ATOM_regen_nv_OpenCodeInstruct

ATOM regen: OpenCodeInstruct responses by Kimi-K3 用 Kimi-K3 对 nvidia/OpenCodeInstruct 的 prompt 重新生成了一遍回答,供 off-policy 投机解码蒸馏(SDDD)使用。 窗口 32768。 为什么是代码,为什么是这个数据集 训练一个 5 层 DSpark draft 的过程里,代码一直是最薄的一片:主语料去重之后 code 只剩 23,083 行唯一 prompt(17.5 万行里平均一题 5.6 个回答,17.9% 是物理 上限),占整个语料的 **0.43%**。对应的后果是 SWE-bench Pro 成为第二弱的 benchmark —— 只有同机官方 draft 的 **84.9%**。 把代码占比从 0.43% 提到 **20%**(主要靠这份数据)之后,SWE-bench Pro 到了 **100.6%**,是 13 套 benchmark 里相对提升最大的一套。 这份数据还顺带解决另一件事:长 prompt… See the full description on the dataset page: https://huggingface.co/datasets/slippedJim/ATOM_regen_nv_OpenCodeInstruct.

sourceHugging Faceapache-2.0updated 6d agoView on Hugging Face
0likes1downloads
README.mdDownload Raw Back to root

This repository is gated, so its file contents are only served once you have accepted the publisher's terms at Hugging Face. Open it at the source above.