Datasetpublic

saintsauce/unified-vlm-steering-eval

Unified VLM steering eval Activation steering of unified vision-language models: steered text and image generations, steering vectors and judge scores. One folder per model; prompts/ is shared. Steering is h <- h + alpha * v_hat everywhere. model run text rows image rows concepts with images published uniar/ av_v2 21,560 21,560 age, chaos, cleanness, emotion, near_far, size, spatial_lr 2026-09-11 Layout per model: manifest.json (exact config)… See the full description on the dataset page: https://huggingface.co/datasets/saintsauce/unified-vlm-steering-eval.

sourceHugging Faceupdated 16d agoView on Hugging Face

0likes211downloads

36 commits on main

ed86eb116d ago

uniar: publish 10/10 (20 files)

saintsauce

6f808c316d ago

uniar: publish 9/10 (160 files)

saintsauce

de630ec16d ago

uniar: publish 8/10 (2500 files)

saintsauce

3270b5516d ago

uniar: publish 7/10 (2500 files)

saintsauce

fc7650316d ago

uniar: publish 6/10 (2500 files)

saintsauce

07da08b16d ago

uniar: publish 5/10 (2500 files)

saintsauce

a26ce9716d ago

uniar: publish 4/10 (2500 files)

saintsauce

6d8b43116d ago

uniar: publish 3/10 (2500 files)

saintsauce

79e2e5f16d ago

uniar: publish 2/10 (2500 files)

saintsauce

e66930f16d ago

uniar: publish 1/10 (2500 files)

saintsauce

6ab89d816d ago

uniar: remove 2495 stale image paths

saintsauce

c35746316d ago

uniar: remove 2500 stale image paths

saintsauce

267741816d ago

uniar: remove 2500 stale image paths

saintsauce

1314c3a16d ago

uniar: remove 2500 stale image paths

saintsauce

4a0392716d ago

uniar: publish 1/10 (20 files)

saintsauce

b179c1016d ago