W-61/qwen3-8b-base-new-dpo-ultrafeedback-4xh200-batch-128-q_t-0.43-s_star-0.35-20260430-192039-margin
收藏资源简介:
该数据集是从一个New-DPO训练运行中导出的每一步边缘(margin)统计摘要。数据集包含了训练过程中的多个特征,如epoch、step、batch_size、mean、std、min、p10、median、p90、max、pos_frac、sample和npy等。数据集的来源运行信息包括模型仓库ID、基础模型、训练运行名称、W&B项目、训练器类型、边缘日志路径等。训练参数包括beta、f_divergence_type、f_alpha_divergence_coef、s_star、eta和q_t等。数据集混合器使用了HuggingFaceH4/ultrafeedback_binarized。
Per-step margin summary statistics exported from a New-DPO training run. The dataset includes multiple features such as epoch, step, batch_size, mean, std, min, p10, median, p90, max, pos_frac, sample, and npy. The source run information includes model repo id, base model, training run name, W&B project, trainer type, margin log path, etc. The training arguments include beta, f_divergence_type, f_alpha_divergence_coef, s_star, eta, and q_t. The dataset mixer uses HuggingFaceH4/ultrafeedback_binarized.




