Upload configuration_spice_cnn.py with huggingface_hub
Browse files- configuration_spice_cnn.py +49 -0
configuration_spice_cnn.py
ADDED
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import PretrainedConfig
|
2 |
+
|
3 |
+
"""Spice CNN model configuration"""
|
4 |
+
|
5 |
+
SPICE_CNN_PRETRAINED_CONFIG_ARCHIVE_MAP = {
|
6 |
+
"spicecloud/spice-cnn-base": "https://huggingface.co/spice-cnn-base/resolve/main/config.json"
|
7 |
+
}
|
8 |
+
|
9 |
+
|
10 |
+
# Define custom convnet configuration
|
11 |
+
class SpiceCNNConfig(PretrainedConfig):
|
12 |
+
"""
|
13 |
+
This is the configuration class to store the configuration of a [`SpiceCNNModel`].
|
14 |
+
It is used to instantiate an SpiceCNN model according to the specified arguments,
|
15 |
+
defining the model architecture. Instantiating a configuration with the defaults
|
16 |
+
will yield a similar configuration to that of the SpiceCNN
|
17 |
+
[spicecloud/spice-cnn-base](https://huggingface.co/spicecloud/spice-cnn-base)
|
18 |
+
architecture.
|
19 |
+
|
20 |
+
Configuration objects inherit from [`PretrainedConfig`] and can be used to control
|
21 |
+
the model outputs. Read the documentation from [`PretrainedConfig`] for more
|
22 |
+
information.
|
23 |
+
"""
|
24 |
+
|
25 |
+
model_type = "spicecnn"
|
26 |
+
|
27 |
+
def __init__(
|
28 |
+
self,
|
29 |
+
in_channels: int = 3,
|
30 |
+
num_classes: int = 10,
|
31 |
+
dropout_rate: float = 0.4,
|
32 |
+
hidden_size: int = 128,
|
33 |
+
num_filters: int = 16,
|
34 |
+
kernel_size: int = 3,
|
35 |
+
stride: int = 1,
|
36 |
+
padding: int = 1,
|
37 |
+
pooling_size: int = 2,
|
38 |
+
**kwargs
|
39 |
+
):
|
40 |
+
super().__init__(**kwargs)
|
41 |
+
self.in_channels = in_channels
|
42 |
+
self.num_classes = num_classes
|
43 |
+
self.dropout_rate = dropout_rate
|
44 |
+
self.hidden_size = hidden_size
|
45 |
+
self.num_filters = num_filters
|
46 |
+
self.kernel_size = kernel_size
|
47 |
+
self.stride = stride
|
48 |
+
self.padding = padding
|
49 |
+
self.pooling_size = pooling_size
|