Update README.md
Browse files
README.md
CHANGED
|
@@ -43,7 +43,7 @@ The no caption dataset uses only ohwx 3d render as caption
|
|
| 43 |
|
| 44 |
<img src="https://cdn-uploads.huggingface.co/production/uploads/6345bd89fe134dfd7a0dba40/jK75d8i1x5hAHSYSsJNBd.png" alt="Training configuration" style="max-height: 500px; width: auto;">
|
| 45 |
|
| 46 |
-
All trainings are saved as Float and 128 LoRA
|
| 47 |
|
| 48 |
## Inconsistent Dataset Training
|
| 49 |
|
|
|
|
| 43 |
|
| 44 |
<img src="https://cdn-uploads.huggingface.co/production/uploads/6345bd89fe134dfd7a0dba40/jK75d8i1x5hAHSYSsJNBd.png" alt="Training configuration" style="max-height: 500px; width: auto;">
|
| 45 |
|
| 46 |
+
All trainings are saved as Float and 128 LoRA Network Rank thus they are above 2GB per checkpoint
|
| 47 |
|
| 48 |
## Inconsistent Dataset Training
|
| 49 |
|