Alignment datasets by statsmind

Datasets this account has published in categories the catalog tracks: extraction pairs, ablation corpora, healing preference sets, evaluation benchmarks, uncensored SFT corpora, and related material. Category badges link to the workflow stage.

2 in /datasets
Dataset
Stage
Downloads
hh_rlhf_cn 0 0
基于dikw/hh_rlhf_cn,对 hh_rlhf_train.jsonl 和 hh_rlhf_test.jsonl 重新进行整理 去掉 context column, 保留 chosen, rejected 两列, 格式为 conversational
226
Dataset
Stage
Downloads
hh_rlhf_cn 0 0
基于dikw/hh_rlhf_cn,对 hh_rlhf_train.jsonl 和 hh_rlhf_test.jsonl 重新进行整理 去掉 context column, 保留 chosen, rejected 两列, 格式为 conversational
226
Dataset
Stage
Downloads
hh_rlhf_cn 0 0
基于dikw/hh_rlhf_cn,对 hh_rlhf_train.jsonl 和 hh_rlhf_test.jsonl 重新进行整理 去掉 context column, 保留 chosen, rejected 两列, 格式为 conversational
226
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration