W-61/llama-3-8b-base-new-dpo-hh-harmless-4xh200-batch-64-s_star-0.4-eta-0.1-q_t-0.48-margin-log

该数据集是从New-DPO训练运行中导出的每步边际摘要统计信息。包含训练过程中的各种统计特征，如epoch、step、batch_size、mean、std等。数据集来源为模型仓库`W-61/llama-3-8b-base-new-dpo-hh-harmless-4xh200-batch-64-s_star-0.4-eta-0.1-q_t-0.48`，基础模型为`W-61/llama-3-8b-base-sft-hh-harmless-4xh200`。训练参数包括beta、f_divergence_type、s_star、eta等。数据集混合器使用了`Anthropic/hh-rlhf`。

This dataset contains per-step margin summary statistics exported from a New-DPO training run. It includes various statistical features during the training process, such as epoch, step, batch_size, mean, std, etc. The dataset originates from the model repository `W-61/llama-3-8b-base-new-dpo-hh-harmless-4xh200-batch-64-s_star-0.4-eta-0.1-q_t-0.48`, with the base model being `W-61/llama-3-8b-base-sft-hh-harmless-4xh200`. Training parameters include beta, f_divergence_type, s_star, eta, etc. The dataset mixer uses `Anthropic/hh-rlhf`.