Alignment datasets by dvilasuero

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.

14 in /datasets
Dataset
Stage
Downloads
jailbreak-classification-reasoning-eval 0 0
Eval models for classification on your own data This dataset contains the results of evaluating reasoning models for classification. It c...
718
intel-orca-dpo-pair-rg 0 0
Dataset Card for intel-orca-dpo-pair-rg This dataset has been created with Argilla. As shown in the sections below, this dataset can be l...
513
jailbreak-classification-reasoning-results 0 0
Results Summary Model Accuracy Total Correct Unparsable qwq32 90.00% 100 90 4 r1 93.00% 100 93 3 llama70B 69.00% 100 69 18...
358
Dataset
Stage
Downloads
jailbreak-classification-reasoning-eval 0 0
Eval models for classification on your own data This dataset contains the results of evaluating reasoning models for classification. It c...
718
jailbreak-classification-reasoning-results 0 0
Results Summary Model Accuracy Total Correct Unparsable qwq32 90.00% 100 90 4 r1 93.00% 100 93 3 llama70B 69.00% 100 69 18...
358
intel-orca-dpo-pair-rg 0 0
Dataset Card for intel-orca-dpo-pair-rg This dataset has been created with Argilla. As shown in the sections below, this dataset can be l...
513
Dataset
Stage
Downloads
jailbreak-classification-reasoning-eval 0 0
Eval models for classification on your own data This dataset contains the results of evaluating reasoning models for classification. It c...
718
intel-orca-dpo-pair-rg 0 0
Dataset Card for intel-orca-dpo-pair-rg This dataset has been created with Argilla. As shown in the sections below, this dataset can be l...
513
jailbreak-classification-reasoning-results 0 0
Results Summary Model Accuracy Total Correct Unparsable qwq32 90.00% 100 90 4 r1 93.00% 100 93 3 llama70B 69.00% 100 69 18...
358
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