Team Ai
Datasetpublic

Rutabin/deception-probing-tutorial-lite

Deception probing tutorial — Gemma-2-9B-IT activations (lite) Precomputed residual-stream activations for a hands-on replication of Natarajan et al. (2026), One Probe Won't Catch Them All (arXiv:2602.01425), which builds on Goldowsky-Dill et al. (2025), Detecting Strategic Deception with Linear Probes. The point of shipping activations rather than a model: everything scientifically interesting in both papers happens downstream of the forward pass. With these vectors the whole… See the full description on the dataset page: https://huggingface.co/datasets/Rutabin/deception-probing-tutorial-lite.

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face
0likes42downloads
revisionmainv1
filealpaca_control.npz13.4 MBdownload
fileapollo_ai_liar.npz1.8 MBdownload
fileapollo_roleplaying.npz10.1 MBdownload
fileapollo_sandbagging.npz13.5 MBdownload
fileinstructed_pairs_pertoken.npz14.5 MBdownload
fileinstructed_pairs.npz26.7 MBdownload
filelietype_concealment.npz13.4 MBdownload
filelietype_evasion.npz13.4 MBdownload
filelietype_exaggeration.npz13.4 MBdownload
filelietype_overt_lie.npz13.4 MBdownload
filelietype_partial_truth.npz13.4 MBdownload
filelietype_pressure_dishonesty.npz13.4 MBdownload
filelietype_white_lie.npz13.4 MBdownload

Rutabin/deception-probing-tutorial-lite · main · files are served by the source, never re-hosted here