Insights DB

Last scanned: Never Scanned

Version	Commit message	Commit author	Last scanned
8dda1e	Re-validate quant on 12 prompts (n=96): honest numbers + updated log	avlp12	Never scanned
e337fe	Pure-MLX NSFW safety filter (torch-free, validated 2e-7 vs PyTorch) + review hardening	avlp12
a85351	Harden per code review: strict loads, input validation, safe downloads, OOM-safe model switch	avlp12
b52fa3	Upload generate.py with huggingface_hub	avlp12
7eba41	Upload app.py with huggingface_hub	avlp12
093f59	Upload krea2/pipeline.py with huggingface_hub	avlp12
146dbb	Upload krea2/safety.py with huggingface_hub	avlp12
5e504a	Upload generate.py with huggingface_hub	avlp12
e9ecdc	Upload app.py with huggingface_hub	avlp12
d53aad	Upload README.md with huggingface_hub	avlp12
4afe6f	Upload krea2/pipeline.py with huggingface_hub	avlp12
4976f7	Upload krea2/sampling.py with huggingface_hub	avlp12
801277	Upload krea2/pipeline.py with huggingface_hub	avlp12
3f287e	Upload app.py with huggingface_hub	avlp12
759992	Upload generate.py with huggingface_hub	avlp12
ff1e63	Upload app.py with huggingface_hub	avlp12
cbb5a8	Upload README.md with huggingface_hub	avlp12
ccfdc0	Upload krea2/pipeline.py with huggingface_hub	avlp12
52f05d	Upload generate.py with huggingface_hub	avlp12
8f5ebb	Upload app.py with huggingface_hub	avlp12

‌
‌
‌
‌
‌
‌
‌
‌
‌
‌

With Prisma AIRS you can scan models for threats before ML developers download them for use, and apply policies based on your risk tolerance.

Earn bounties and get recognition for your discoveries by submitting them through Huntr.

Products and Services