Alignment datasets by ZeLi111

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.

1 in /datasets
Dataset
Stage
Downloads
StupidPolicemanOnlyKnowReject-Chinese-uncensored 0 0
总所周知,大语言模型都会拒绝用户,无论你的问题是否合法,一些人通过搜集这些大模型的回答用于二次训练自己的模型,但是问题是这些预料中包含了大量的拒绝模板,导致模型性能大打折扣,因为模型没有学会任何知识,模型此时只是一只看门狗,一个没有任何文化的警察,因为模型只在乎合规,只在乎如何拒...
908
Dataset
Stage
Downloads
StupidPolicemanOnlyKnowReject-Chinese-uncensored 0 0
总所周知,大语言模型都会拒绝用户,无论你的问题是否合法,一些人通过搜集这些大模型的回答用于二次训练自己的模型,但是问题是这些预料中包含了大量的拒绝模板,导致模型性能大打折扣,因为模型没有学会任何知识,模型此时只是一只看门狗,一个没有任何文化的警察,因为模型只在乎合规,只在乎如何拒...
908
Dataset
Stage
Downloads
StupidPolicemanOnlyKnowReject-Chinese-uncensored 0 0
总所周知,大语言模型都会拒绝用户,无论你的问题是否合法,一些人通过搜集这些大模型的回答用于二次训练自己的模型,但是问题是这些预料中包含了大量的拒绝模板,导致模型性能大打折扣,因为模型没有学会任何知识,模型此时只是一只看门狗,一个没有任何文化的警察,因为模型只在乎合规,只在乎如何拒...
908
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