Alignment datasets by kenhktsui

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
test_toxic_chat just added 0
Please note that this is subset of lmsys/toxic-chat annotated by anyclassifier.This dataset may contain racism, sexuality, or other undesire...
585
test_toxic_chat_syn just added 0
Please note that this is synthetic data generated by anyclassifier, so it is not real.This dataset may contain racism, sexuality, or other u...
545
Dataset
Stage
Downloads
test_toxic_chat just added 0
Please note that this is subset of lmsys/toxic-chat annotated by anyclassifier.This dataset may contain racism, sexuality, or other undesire...
585
test_toxic_chat_syn just added 0
Please note that this is synthetic data generated by anyclassifier, so it is not real.This dataset may contain racism, sexuality, or other u...
545
Dataset
Stage
Downloads
test_toxic_chat just added 0
Please note that this is subset of lmsys/toxic-chat annotated by anyclassifier.This dataset may contain racism, sexuality, or other undesire...
585
test_toxic_chat_syn just added 0
Please note that this is synthetic data generated by anyclassifier, so it is not real.This dataset may contain racism, sexuality, or other u...
545
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