Commit History
2 days ago
-
Add an 8-bit vision model with int8 compute (160 px)
Verified signature9c04489
15 days ago
-
imagevolve: model card rewritten, demo video of the web interface
Verified signatureRenamed from apophenia-clip-vit-base-patch32. The files are unchanged; imagevolve loads revision 98c33fd. demo.mp4 is AV1, demo-h264.mp4 its H.264 fallback.8d9f029
16 days ago
-
A vision model for 160-pixel input, half the work per picture
Verified signature98c33fd -
CLIP ViT-B/32 q4f16 for apophenia, the text model's token table in 8 bits
Verified signaturee16ed39