comfyanonymous/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-02-24 17:04:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

[Trainer] training with proper offloading (#12189)

Browse Source 
* Fix bypass dtype/device moving

* Force offloading mode for training

* training context var

* offloading implementation in training node

* fix wrong input type

* Support bypass load lora model, correct adapter/offloading handling
This commit is contained in:
Kohaku-Blueleaf
2026-02-11 10:45:19 +08:00
committed by GitHub
parent dbe70b6821
commit cdcf4119b3
5 changed files with 196 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
comfy/model_management.py
View File

@@ -55,6 +55,11 @@ cpu_state = CPUState.GPU
total_vram = 0
# Training Related State
in_training = False
def get_supported_float8_types():
float8_types = []
try: