You can deploy this model and start earning money today!

guanmingchiu/
lrcu-dream-7b-wmdp-cyber

Name: lrcu-dream-7b-wmdp-cyber
Availability: InStock

safetensors

Dream

unlearning

diffusion-language-model

safety

custom_code

Dream-org/Dream-v0-Instruct-7B

finetune

mit

Views

No views yet

LRCU-unlearned Dream-v0-Instruct-7B (WMDP-cyber)

Dream-org/Dream-v0-Instruct-7B unlearned on WMDP-cyber with LRCU (Localized Recall-Capped Unlearning): a saturating per-token recall cap on a causally-localized block band (blocks 9-11) + bounded Min-SNR (low-$t$) trajectory weighting + CE retain anchor, with no reference model.

python

from transformers import AutoModel, AutoTokenizer
m = AutoModel.from_pretrained("guanmingchiu/lrcu-dream-7b-wmdp-cyber", trust_remote_code=True)
t = AutoTokenizer.from_pretrained("guanmingchiu/lrcu-dream-7b-wmdp-cyber", trust_remote_code=True)