tensorflow获取ckpt的参数

梦里梦外; 2023-08-17 17:04 94阅读 0赞

import tensorflow as tf
    from tensorflow.python import pywrap_tensorflow
    from tensorflow.python.tools.inspect_checkpoint import print_tensors_in_checkpoint_file

方法1， 采用print\_tensors\_in\_checkpoint\_file 输出全部的参数

model_dir = "tfPara/"
     
    ckpt = tf.train.get_checkpoint_state(model_dir)
    ckpt_path = ckpt.model_checkpoint_path
    print_tensors_in_checkpoint_file(ckpt_path, all_tensors=True,all_tensor_names = True, tensor_name='')

tensor_name:  beta1_power
    0.3138105
    tensor_name:  beta2_power
    0.989055
    tensor_name:  l1/bias
    [ 0.02627242  0.05014378 -0.04511173  0.        ...
    tensor_name:  l1/bias/Adam
    [-3.4903574e+00 -2.6453564e+01  6.5189457e+00 ...
    tensor_name:  l1/bias/Adam_1
    [3.0681181e-01 1.6404903e+01 1.7573323e+00 ...
    tensor_name:  l1/kernel
    [[-0.05044619  0.07713371 -0.3241992   0.02499923  0.04578751 -0.2615546
       0.16445747 -0.34566906  0.2644527  -0.24029821 -0.1680427  -0.31824082
       0.2339457  -0.13366055  0.30801657  0.3266713  -0.3441043  -0.16687115
       0.06713113  0.22774872] ...

方法2 输出指定层的指定参数

其中，LAYER\_1\_NAME = 'l1’就是在用 tf.layers.dense() 创建网络的一个layer的时候（l1 = tf.layers.dense(tf\_x, 20, tf.nn.relu, name =‘l1’)）

这里的kernel 指的是权重参数，bias指的是bias

参考链接：  
https://www.codelast.com/%e5%8e%9f%e5%88%9b-%e5%a6%82%e4%bd%95%e5%8f%96%e5%87%ba-tf-layers-dense-%e5%ae%9a%e4%b9%89%e7%9a%84%e5%85%a8%e8%bf%9e%e6%8e%a5%e5%b1%82%e7%9a%84weight%e5%92%8cbias%e5%8f%82%e6%95%b0%e5%80%bc/

model_dir = "tfPara/"
     
    ckpt = tf.train.get_checkpoint_state(model_dir)
    ckpt_path = ckpt.model_checkpoint_path
    reader = tf.train.NewCheckpointReader(ckpt_path)
    
    file = 'tfPara/para.txt'
    layer_names = ['l1','l2','l3']
    with open(file, 'w') as f:
        
        for ln in layer_names:
            weights = reader.get_tensor(ln + '/kernel')  # weight的名字，是由对应层的名字，加上默认的"kernel"组成的
            bias = reader.get_tensor(ln + '/bias')  # bias的名字
            print(weights)
            print(bias)
            f.write(ln + '/weights; shape:'+str(weights.shape) + '\n')
            f.write(str(weights.tolist()))
            f.write('\n')
            f.write(ln + '/bias; shape:'+str(bias.shape)+ '\n')
            f.write(str(bias.tolist()))
            f.write('\n')

[[-0.05044619  0.07713371 -0.3241992   0.02499923  0.04578751 -0.2615546
       0.16445747 -0.34566906  0.2644527  -0.24029821 -0.1680427  -0.31824082
       0.2339457  -0.13366055  0.30801657  0.3266713  -0.3441043  -0.16687115
       0.06713113  0.22774872] ...
    
    [[ 0.22894777]
     [ 0.49630672] ...