Commit
•
93de90d
1
Parent(s):
f11042f
Update README.md
Browse files
README.md
CHANGED
|
@@ -2,13 +2,13 @@ Fashion MNIST unconditional Unet Model trained using DDPM
|
|
| 2 |
|
| 3 |
Model Hyperparams:
|
| 4 |
|
| 5 |
-
- Model size:
|
| 6 |
- 3 stages: 128, 256, 512 channels
|
| 7 |
- Linear Attention in 2nd and 3rd stages, Self Attention in Middle Stage
|
| 8 |
- Optimizer: Adam
|
| 9 |
- LR: 3e-4
|
| 10 |
- Batch Size: 64
|
| 11 |
- Grad Accumulation: 8 steps
|
| 12 |
-
|
| 13 |
|
| 14 |
|
|
|
|
| 2 |
|
| 3 |
Model Hyperparams:
|
| 4 |
|
| 5 |
+
- Model size: 51,834,625 params
|
| 6 |
- 3 stages: 128, 256, 512 channels
|
| 7 |
- Linear Attention in 2nd and 3rd stages, Self Attention in Middle Stage
|
| 8 |
- Optimizer: Adam
|
| 9 |
- LR: 3e-4
|
| 10 |
- Batch Size: 64
|
| 11 |
- Grad Accumulation: 8 steps
|
| 12 |
+
- Total steps: 5,000
|
| 13 |
|
| 14 |
|