localbitsback

localbitsback copied to clipboard

Finetuning

1

May I ask is there any possibility that I can fine-tune models for my own dataset? If there is, how can I do that?

DaniMlk

```python import torch from compression.models.load_flowpp_imagenet64 import load_imagenet64_model model = load_imagenet64_model('/your/data/folder/flowpp_imagenet64_model.npz', False) noise = torch.randn(1, 192, 8, 8).double() img, _ = model.main_flow(noise, aux=None, inverse=True) noise2, _ = model.main_flow(img, aux=None, inverse=False) ```...

atgambardella