datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
OpenHermes-NoRefusal-95K
OpenHermes-NoRefusal-95K
A refusal-free instruction-tuning dataset: 95,401 single-turn conversations derived from
teknium/OpenHermes-2.5, filtered so
that zero assistant responses contain refusals, hedging boilerplate, or
"as an AI language model" disclaimers.
Why this exists
The usual way to get a model that doesn't refuse is to train it on aligned data and then
remove the alignment afterwards — refusal-direction ablation, weight editing, abliteration.
That works… See the full description on the dataset page: https://huggingface.co/datasets/ghost-actual/OpenHermes-NoRefusal-95K.cybersec-fact-recall
Cybersec Fact-Recall Benchmark (GhostLM v2)
Free-form short-answer benchmark for small cybersecurity language
models. Built and used by the GhostLM
project as the truth metric for the ghost-base v1.0 acceptance gate.
Why this exists
Multiple-choice cybersec benchmarks like CTIBench and SecQA reward
register matching (the model picks the option that "looks like" a
security answer) as much as actual factual recall. A small from-
scratch model can hit 28-30% on those without… See the full description on the dataset page: https://huggingface.co/datasets/Ghostgim/cybersec-fact-recall.