ceselder/loracle-ia-merged-ws-rl
收藏Hugging Face2026-04-27 更新2026-05-03 收录
下载链接:
https://hf-mirror.com/datasets/ceselder/loracle-ia-merged-ws-rl
下载链接
链接失效反馈官方服务:
资源简介:
该数据集名为LoRAcle IA合并warmstart+RL池,是由ceselder/loracle-ia-warmstart和ceselder/loracle-ia-RL两个数据集合并而成。经过过滤,移除了20个特定的fair-eval heldout backdoor orgs。数据集包含2595行数据,涉及950个独特的LoRAs,数据来源包括ia、dpo_pretrain和pretrain_dpo_heldout。建议用作单一的SFT/RL训练池。
The dataset is named LoRAcle IA merged warmstart+RL pool, which is merged from two datasets: ceselder/loracle-ia-warmstart and ceselder/loracle-ia-RL. It has been filtered to remove 20 specific fair-eval heldout backdoor orgs. The dataset contains 2595 rows, involving 950 unique LoRAs, with data sources including ia, dpo_pretrain, and pretrain_dpo_heldout. It is recommended to be used as a single SFT/RL training pool.
提供机构:
ceselder



