Python实现提取.h5模型每层的参数并保存为.bin文件

好的，以下是Python实现提取.h5模型每层的参数并保存为.bin文件的代码示例：

import h5py
import numpy as np

def save_weights_to_binary(weights_file_path):
    with h5py.File(weights_file_path, mode='r') as f:
        # 遍历模型每一层
        for layer_name in f.keys():
            layer_group = f[layer_name]
            # 保存每层的权重参数
            for param_name in layer_group.keys():
                param = layer_group[param_name][...]
                param_file_name = f'{layer_name}_{param_name}.bin'
                np.array(param).astype(np.float32).tofile(param_file_name)

if __name__ == '__main__':
    # 模型权重文件路径
    weights_file_path = 'model_weights.h5'
    save_weights_to_binary(weights_file_path)

运行代码后，会遍历模型的每一层，提取每层的参数并将其保存为对应的二进制文件，例如：

conv1_kernel.bin
conv1_bias.bin
conv2_kernel.bin
conv2_bias.bin
...

每个参数文件包含该参数的所有值，存储为二进制形式，并以float32类型存储。这些二进制文件可以方便地在C++等其他语言中加载使用。