 viralvfx
|
970e85ea02
|
Fix static noise bug by dequantizing FP8 weights before INT4 quantization
|
2026-07-10 17:30:57 +05:30 |
|
|
viralvfx
|
d38dd4520c
|
Fix: Support FP8 weights deferred on-the-fly quantization under AIMDO dynamic staging
v1.0.5
|
2026-07-10 16:42:51 +05:30 |
|
|
viralvfx
|
df8197ebb3
|
Fix: Add missing QuantizedTensor import inside custom_state_dict_for_saving to resolve NameError
v1.0.4
|
2026-07-10 16:34:11 +05:30 |
|
|
viralvfx
|
5f1667705b
|
Fix: Intercept state_dict_for_saving on ModelPatcher to correctly unpack QuantizedTensors during save
v1.0.3
|
2026-07-10 16:28:54 +05:30 |
|
|
viralvfx
|
bd46f24974
|
Fix: Resolve Windows invalid buffer ERROR_INVALID_USER_BUFFER (os error 1784) during safetensors serialization
v1.0.2
|
2026-07-10 16:25:16 +05:30 |
|
|
viralvfx
|
e8bc904f66
|
Chore: Add pyproject.toml configuration and release version 1.0.1
v1.0.1
|
2026-07-10 16:18:43 +05:30 |
|
|
viralvfx
|
12fee3d270
|
Feat: Add support for FP8 weights on-the-fly quantization
|
2026-07-10 16:17:49 +05:30 |
|
|
viralvfx
|
80a65c4742
|
Doc: Add ComfyUI latest version requirement in README
|
2026-07-10 13:01:39 +05:30 |
|
|
viralvfx
|
39879ff2ed
|
Doc: Update installation guide with git clone command in README
|
2026-07-10 12:59:45 +05:30 |
|
|
viralvfx
|
05d0f9baea
|
Doc: Add note about first-run compilation in README
|
2026-07-10 12:58:05 +05:30 |
|
|
viralvfx
|
ae53fbbace
|
Doc: Add Krea2 Turbo INT4 verified model info in README
|
2026-07-10 12:56:00 +05:30 |
|
|
viralvfx
|
850cec79ef
|
Initial commit: Standalone ComfyUI INT4 fast inference custom node
|
2026-07-10 12:50:59 +05:30 |
|