Hugging Face's logo

lilcheaty
/
Krea2-INT8-ConvRot

Text-to-Image
krea
krea2
int8
convrot
comfyui
quantized
Model card Files
xet
 Community
3
lilcheaty commited on
Commit
030fc9d
·
verified ·
1 Parent(s): 194b08b

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -30,8 +30,8 @@ tensor-core acceleration on Ampere) and roughly half the VRAM of BF16.
30
 
31
  | File | Precision | Size | Use |
32
  |------|-----------|------|-----|
33
- | `Krea2-Turbo-int8-ConvRot.safetensors` | INT8 ConvRot | 13.2 GB | 8-step distilled, fast text-to-image |
34
- | `Krea2-Raw-int8-ConvRot.safetensors` | INT8 ConvRot | ~13 GB | Undistilled base — fine-tuning / LoRA training / research |
35
 
36
  Original BF16 checkpoints are ~26.6 GB each. Quantized with the K2 profile: the 28 main DiT blocks are INT8,
37
  while the sensitive layers (`first`, `last`, `tmlp`, `tproj`, `txtfusion`, `txtmlp`) are kept in high precision.
 
30
 
31
  | File | Precision | Size | Use |
32
  |------|-----------|------|-----|
33
+ | `Krea2-Turbo-int8-ConvRot.safetensors` | INT8 ConvRot | 14.1 GB | 8-step distilled, fast text-to-image |
34
+ | `Krea2-Raw-int8-ConvRot.safetensors` | INT8 ConvRot | 14.1 GB | Undistilled base — fine-tuning / LoRA training / research |
35
 
36
  Original BF16 checkpoints are ~26.6 GB each. Quantized with the K2 profile: the 28 main DiT blocks are INT8,
37
  while the sensitive layers (`first`, `last`, `tmlp`, `tproj`, `txtfusion`, `txtmlp`) are kept in high precision.