Update layout for checkpoints/ prefix
Browse files
README.md
CHANGED
|
@@ -69,10 +69,12 @@ trained with, so the table above can be checked rather than trusted.
|
|
| 69 |
## Repository layout
|
| 70 |
|
| 71 |
```
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
|
|
|
|
|
|
| 76 |
samples/
|
| 77 |
base/ vanilla SD 1.5, the comparison baseline
|
| 78 |
checkpoint-<step>/
|
|
@@ -102,7 +104,7 @@ pipe = StableDiffusionPipeline.from_pretrained(
|
|
| 102 |
"runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16
|
| 103 |
).to("cuda")
|
| 104 |
pipe.load_lora_weights(
|
| 105 |
-
"whosouravsharma/diffusiondb-sd15-lora", subfolder="checkpoint-
|
| 106 |
)
|
| 107 |
|
| 108 |
image = pipe(
|
|
|
|
| 69 |
## Repository layout
|
| 70 |
|
| 71 |
```
|
| 72 |
+
checkpoints/
|
| 73 |
+
checkpoint-<step>/
|
| 74 |
+
pytorch_lora_weights.safetensors the adapter
|
| 75 |
+
optimizer.pt optimizer + scaler state, for resuming
|
| 76 |
+
state.json step, epoch, hyperparameters
|
| 77 |
+
training/ the scripts that produced all of this
|
| 78 |
samples/
|
| 79 |
base/ vanilla SD 1.5, the comparison baseline
|
| 80 |
checkpoint-<step>/
|
|
|
|
| 104 |
"runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16
|
| 105 |
).to("cuda")
|
| 106 |
pipe.load_lora_weights(
|
| 107 |
+
"whosouravsharma/diffusiondb-sd15-lora", subfolder="checkpoints/checkpoint-4240"
|
| 108 |
)
|
| 109 |
|
| 110 |
image = pipe(
|