Alignment datasets by xsample

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.

8 in /datasets
Dataset
Stage
Downloads
tulu-3-mig-50k 0 0
Tulu-3-MIG-50K Project | Github | Paper | HuggingFace's collection MIG is an automatic data selection method for instruction tuning. This...
1.4k
tulu-3-qdit-50k 0 0
Tulu-3-QDIT-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and d...
595
tulu-3-ifd-50k 0 0
Tulu-3-IFD-50K Project | Github | Paper | HuggingFace's collection This dataset includes 50K high-quality SFT data sampled from Tulu3 usi...
477
tulu-3-deita-50k 0 0
Tulu-3-DEITA-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and ...
455
tulu-3-car-50k 0 0
Tulu-3-CaR-50K Project | Github | Paper | HuggingFace's collection This dataset includes 50K high-quality and diverse SFT data sampled fr...
408
tulu-3-random-50k 0 0
Tulu-3-Random-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K SFT data randoml...
390
tulu-3-zip-50k 0 0
Tulu-3-ZIP-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K diverse SFT data sa...
386
tulu-3-instag-50k 0 0
Tulu-3-InsTag-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and...
366
Dataset
Stage
Downloads
tulu-3-deita-50k 0 0
Tulu-3-DEITA-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and ...
455
tulu-3-zip-50k 0 0
Tulu-3-ZIP-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K diverse SFT data sa...
386
tulu-3-random-50k 0 0
Tulu-3-Random-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K SFT data randoml...
390
tulu-3-qdit-50k 0 0
Tulu-3-QDIT-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and d...
595
tulu-3-instag-50k 0 0
Tulu-3-InsTag-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and...
366
tulu-3-ifd-50k 0 0
Tulu-3-IFD-50K Project | Github | Paper | HuggingFace's collection This dataset includes 50K high-quality SFT data sampled from Tulu3 usi...
477
tulu-3-car-50k 0 0
Tulu-3-CaR-50K Project | Github | Paper | HuggingFace's collection This dataset includes 50K high-quality and diverse SFT data sampled fr...
408
tulu-3-mig-50k 0 0
Tulu-3-MIG-50K Project | Github | Paper | HuggingFace's collection MIG is an automatic data selection method for instruction tuning. This...
1.4k
Dataset
Stage
Downloads
tulu-3-mig-50k 0 0
Tulu-3-MIG-50K Project | Github | Paper | HuggingFace's collection MIG is an automatic data selection method for instruction tuning. This...
1.4k
tulu-3-deita-50k 0 0
Tulu-3-DEITA-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and ...
455
tulu-3-qdit-50k 0 0
Tulu-3-QDIT-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and d...
595
tulu-3-ifd-50k 0 0
Tulu-3-IFD-50K Project | Github | Paper | HuggingFace's collection This dataset includes 50K high-quality SFT data sampled from Tulu3 usi...
477
tulu-3-car-50k 0 0
Tulu-3-CaR-50K Project | Github | Paper | HuggingFace's collection This dataset includes 50K high-quality and diverse SFT data sampled fr...
408
tulu-3-random-50k 0 0
Tulu-3-Random-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K SFT data randoml...
390
tulu-3-zip-50k 0 0
Tulu-3-ZIP-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K diverse SFT data sa...
386
tulu-3-instag-50k 0 0
Tulu-3-InsTag-50K Project | Github | Paper | HuggingFace's collection This dataset is a baseline of MIG. It includes 50K high-quality and...
366
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