Alignment datasets by qikp

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-singles 0 0
hh-rlhf-singles This is a collated version of Anthropic/hh-rlhf to convert each row into single prompt/chosen/rejected rows.
211
aninsthro 0 0
Aninsthro Aninsthro is a dataset based on Anthropic/hh-rlhf, specifically the helpful-base portion and the chosen column. This can be use...
112
Dataset
Stage
Downloads
aninsthro 0 0
Aninsthro Aninsthro is a dataset based on Anthropic/hh-rlhf, specifically the helpful-base portion and the chosen column. This can be use...
112
hh-rlhf-singles 0 0
hh-rlhf-singles This is a collated version of Anthropic/hh-rlhf to convert each row into single prompt/chosen/rejected rows.
211
Dataset
Stage
Downloads
hh-rlhf-singles 0 0
hh-rlhf-singles This is a collated version of Anthropic/hh-rlhf to convert each row into single prompt/chosen/rejected rows.
211
aninsthro 0 0
Aninsthro Aninsthro is a dataset based on Anthropic/hh-rlhf, specifically the helpful-base portion and the chosen column. This can be use...
112
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