Jorgvt commited on
Commit
3ad5fbd
·
1 Parent(s): e53b63b

Update README.md

Browse files

Added usage instructions to load the model from the Hub.

Files changed (1) hide show
  1. README.md +16 -1
README.md CHANGED
@@ -36,6 +36,10 @@ Link to the run: https://wandb.ai/jorgvt/PerceptNet/runs/28m2cnzj?workspace=user
36
 
37
  # Usage
38
 
 
 
 
 
39
  As of now to use the model you have to install the [PerceptNet repo](https://github.com/Jorgvt/perceptnet) to get access to the `PerceptNet` class where you will load the weights available here like this:
40
 
41
  ```python
@@ -49,4 +53,15 @@ model.build(input_shape=(None, 384, 512, 3))
49
  model.load_weights(weights_path)
50
  ```
51
 
52
- > PerceptNet requires `wandb` to be installed. It's something we're looking into.
 
 
 
 
 
 
 
 
 
 
 
 
36
 
37
  # Usage
38
 
39
+ There are two alternatives to use the model: install our development repo and load the pretrained weights manually, and load the model using `from_pretrained_keras`:
40
+
41
+ ## Loading weights manually
42
+
43
  As of now to use the model you have to install the [PerceptNet repo](https://github.com/Jorgvt/perceptnet) to get access to the `PerceptNet` class where you will load the weights available here like this:
44
 
45
  ```python
 
53
  model.load_weights(weights_path)
54
  ```
55
 
56
+ > PerceptNet requires `wandb` to be installed. It's something we're looking into.
57
+
58
+ ## Directly from the Hub
59
+
60
+ As every other *Keras* model in the Hub, it can be loaded as follows:
61
+
62
+ ```python
63
+ from huggingface_hub import from_pretrained_keras
64
+ model = from_pretrained_keras("Jorgvt/PerceptNet", compile=False)
65
+ ```
66
+
67
+ > Keep in mind that the model uses grouped convolutions and, at least in Colab, `Unimplemented Errors` may arise when using it in CPU.