W-61/llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.45-s_star-0.4-eta-0.3-margin-log
收藏Hugging Face2026-04-28 更新2026-05-03 收录
下载链接:
https://hf-mirror.com/datasets/W-61/llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.45-s_star-0.4-eta-0.3-margin-log
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是从New-DPO训练运行中导出的每步边缘摘要统计信息。数据集包含多个特征,如epoch(训练轮数)、step(步骤)、batch_size(批次大小)、mean(平均值)、std(标准差)、min(最小值)、p10(第10百分位数)、median(中位数)、p90(第90百分位数)、max(最大值)、pos_frac(正分数)、sample(每个示例的边缘)和npy(保存完整边缘数组的可选路径)。数据集来源于模型库ID为W-61/llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.45-s_star-0.4-eta-0.3的训练运行,使用了Anthropic/hh-rlhf数据集进行混合。
Per-step margin summary statistics exported from a New-DPO training run. The dataset includes features such as epoch, step, batch_size, mean, std, min, p10, median, p90, max, pos_frac, sample (per-example margins for the effective batch on that logged step), and npy (optional path to the saved full margin array). The dataset originates from a training run with model repo id W-61/llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.45-s_star-0.4-eta-0.3 and uses the Anthropic/hh-rlhf dataset for mixing.
提供机构:
W-61



