Integrate mobilenetv4

CI/check_model.py

@@ -1,9 +1,9 @@
 import argparse
-import os
 
+import akida
 from cnn2snn import convert
 from quantizeml import load_model
-from compute_device import compute_min_device
+
 
 def process_model(file_path):
     try:
@@ -18,7 +18,7 @@ def process_model(file_path):
         raise RuntimeError(f"❌ Error converting {file_path}: {e}") from e
 
     try:
-        device = compute_min_device(model_ak, enable_hwpr=True)
+        device = akida.compute_min_device(model_ak, enable_hwpr=True)
         result = len(device.mesh.nps) // 4
         print(f"✅ {file_path}: needs {result} Akida nodes")
     except Exception as e:
@@ -28,10 +28,9 @@ def process_model(file_path):
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
     parser.add_argument("--models", nargs="+", required=True,
-        help="Path to model files (.h5 or .onnx)")
+                        help="Path to model files (.h5 or .onnx)")
     args = parser.parse_args()
 
     # Process each model
     for model_file in args.models:
-        if os.path.exists(model_file):
-            process_model(model_file)
+        process_model(model_file)

README.md

@@ -24,6 +24,7 @@ In addition, some models can be evaluated directly through [Akida Cloud](https:/
 | Vision        | Classification    | [MobileNetV2 1.0](vision/classification/mobilenetv2/imagenet)             | 224        | ImageNet      | 3.5M    | 8            | 70.35%   |          |       | 7              |
 | Vision        | Classification    | [MobileNetV2 0.75](vision/classification/mobilenetv2/imagenet)            | 160        | ImageNet      | 2.6M    | 8            | 62.85%   |          |       | 4 ☁️          |
 | Vision        | Classification    | [MobileNetV2 0.35](vision/classification/mobilenetv2/imagenet)            | 96         | ImageNet      | 1.2M    | 8            | 43.47%   |          |       | 2 ☁️          |
+| Vision        | Classification    | [MobileNetV4 1.0](vision/classification/mobilenetv4/imagenet)             | 224        | ImageNet      | 3.77M   | 8            | 71.86%   |          |       | 8              |
 | Vision        | Classification    | [MobileNetV2_1.0](vision/classification/mobilenetv2/cifar10)              | 128        | CIFAR-10      | 2.25M   | 8            | 93.96%   |          |       | 5 ☁️          |
 | Vision        | Classification    | [MobileNetV2_1.0](vision/classification/mobilenetv2/oxford_flowers)       | 224        | Oxford_Flower | 2.4M    | 8            | 91.97%   |          |       | 7              |
 | Vision        | Classification    | [MobileNetV4_1.0](vision/classification/mobilenetv4/cifar10)              | 128        | CIFAR-10      | 2.5M    | 8            | 94.72%   |          |       | 7              |

vision/classification/mobilenetv4/imagenet/README.md

@@ -0,0 +1,19 @@
+# MobileNetV4-Conv-Small
+
+## Source
+https://huggingface.co/timm/mobilenetv4_conv_small.e2400_r224_in1k
+
+## Environment
+```
+torch: 2.0
+torchvision
+timm
+quantizeml: 1.1.1
+```
+
+## References
+- **MobileNet-v4** Model from the paper [Exploring the MobileNetV4 Architecture](https://arxiv.org/abs/2404.10518)
+- Model Card [MobileNetV4-Conv-Small](https://huggingface.co/timm/mobilenetv4_conv_small.e2400_r224_in1k)
+
+## License
+Please refer to [https://huggingface.co/timm/mobilenetv4_conv_small.e2400_r224_in1k](https://huggingface.co/timm/mobilenetv4_conv_small.e2400_r224_in1k)