Error message running tf_to_keras notebook

[ramiro-feria-puron]

When running the last cell of tf_to_keras notebook (changing model folder and checkpoint to the latest model available), I get the following error:

Loading numpy weights from ../model/keras/npy_weights/
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-14-903bce1a4b25> in <module>
      8             weight_arr = np.load(os.path.join(npy_weights_dir, weight_file))
      9             weights.append(weight_arr)
---> 10         layer.set_weights(weights)
     11 
     12 print('Saving weights...')

~/anaconda3/envs/venv/lib/python3.6/site-packages/keras/engine/base_layer.py in set_weights(self, weights)
   1055                                  str(pv.shape) +
   1056                                  ' not compatible with '
-> 1057                                  'provided weight shape ' + str(w.shape))
   1058             weight_value_tuples.append((p, w))
   1059         K.batch_set_value(weight_value_tuples)

ValueError: Layer weight shape (1792, 128) not compatible with provided weight shape (1792, 512)

Any insight would be really appreciated...

[xiezhuangping]

Same problem.

`Loading numpy weights from ../model/keras/npy_weights/ Conv2d_1a_3x3 Conv2d_1a_3x3_BatchNorm Conv2d_2a_3x3 Conv2d_2a_3x3_BatchNorm Conv2d_2b_3x3 Conv2d_2b_3x3_BatchNorm Conv2d_3b_1x1 Conv2d_3b_1x1_BatchNorm Conv2d_4a_3x3 Conv2d_4a_3x3_BatchNorm Conv2d_4b_3x3 Conv2d_4b_3x3_BatchNorm Block35_1_Branch_2_Conv2d_0a_1x1 Block35_1_Branch_2_Conv2d_0a_1x1_BatchNorm Block35_1_Branch_1_Conv2d_0a_1x1 Block35_1_Branch_2_Conv2d_0b_3x3 Block35_1_Branch_1_Conv2d_0a_1x1_BatchNorm Block35_1_Branch_2_Conv2d_0b_3x3_BatchNorm Block35_1_Branch_0_Conv2d_1x1 Block35_1_Branch_1_Conv2d_0b_3x3 Block35_1_Branch_2_Conv2d_0c_3x3 Block35_1_Branch_0_Conv2d_1x1_BatchNorm Block35_1_Branch_1_Conv2d_0b_3x3_BatchNorm Block35_1_Branch_2_Conv2d_0c_3x3_BatchNorm Block35_1_Conv2d_1x1 Block35_2_Branch_2_Conv2d_0a_1x1 Block35_2_Branch_2_Conv2d_0a_1x1_BatchNorm Block35_2_Branch_1_Conv2d_0a_1x1 Block35_2_Branch_2_Conv2d_0b_3x3 Block35_2_Branch_1_Conv2d_0a_1x1_BatchNorm Block35_2_Branch_2_Conv2d_0b_3x3_BatchNorm Block35_2_Branch_0_Conv2d_1x1 Block35_2_Branch_1_Conv2d_0b_3x3 Block35_2_Branch_2_Conv2d_0c_3x3 Block35_2_Branch_0_Conv2d_1x1_BatchNorm Block35_2_Branch_1_Conv2d_0b_3x3_BatchNorm Block35_2_Branch_2_Conv2d_0c_3x3_BatchNorm Block35_2_Conv2d_1x1 Block35_3_Branch_2_Conv2d_0a_1x1 Block35_3_Branch_2_Conv2d_0a_1x1_BatchNorm Block35_3_Branch_1_Conv2d_0a_1x1 Block35_3_Branch_2_Conv2d_0b_3x3 Block35_3_Branch_1_Conv2d_0a_1x1_BatchNorm Block35_3_Branch_2_Conv2d_0b_3x3_BatchNorm Block35_3_Branch_0_Conv2d_1x1 Block35_3_Branch_1_Conv2d_0b_3x3 Block35_3_Branch_2_Conv2d_0c_3x3 Block35_3_Branch_0_Conv2d_1x1_BatchNorm Block35_3_Branch_1_Conv2d_0b_3x3_BatchNorm Block35_3_Branch_2_Conv2d_0c_3x3_BatchNorm Block35_3_Conv2d_1x1 Block35_4_Branch_2_Conv2d_0a_1x1 Block35_4_Branch_2_Conv2d_0a_1x1_BatchNorm Block35_4_Branch_1_Conv2d_0a_1x1 Block35_4_Branch_2_Conv2d_0b_3x3 Block35_4_Branch_1_Conv2d_0a_1x1_BatchNorm Block35_4_Branch_2_Conv2d_0b_3x3_BatchNorm Block35_4_Branch_0_Conv2d_1x1 Block35_4_Branch_1_Conv2d_0b_3x3 Block35_4_Branch_2_Conv2d_0c_3x3 Block35_4_Branch_0_Conv2d_1x1_BatchNorm Block35_4_Branch_1_Conv2d_0b_3x3_BatchNorm Block35_4_Branch_2_Conv2d_0c_3x3_BatchNorm Block35_4_Conv2d_1x1 Block35_5_Branch_2_Conv2d_0a_1x1 Block35_5_Branch_2_Conv2d_0a_1x1_BatchNorm Block35_5_Branch_1_Conv2d_0a_1x1 Block35_5_Branch_2_Conv2d_0b_3x3 Block35_5_Branch_1_Conv2d_0a_1x1_BatchNorm Block35_5_Branch_2_Conv2d_0b_3x3_BatchNorm Block35_5_Branch_0_Conv2d_1x1 Block35_5_Branch_1_Conv2d_0b_3x3 Block35_5_Branch_2_Conv2d_0c_3x3 Block35_5_Branch_0_Conv2d_1x1_BatchNorm Block35_5_Branch_1_Conv2d_0b_3x3_BatchNorm Block35_5_Branch_2_Conv2d_0c_3x3_BatchNorm Block35_5_Conv2d_1x1 Mixed_6a_Branch_1_Conv2d_0a_1x1 Mixed_6a_Branch_1_Conv2d_0a_1x1_BatchNorm Mixed_6a_Branch_1_Conv2d_0b_3x3 Mixed_6a_Branch_1_Conv2d_0b_3x3_BatchNorm Mixed_6a_Branch_0_Conv2d_1a_3x3 Mixed_6a_Branch_1_Conv2d_1a_3x3 Mixed_6a_Branch_0_Conv2d_1a_3x3_BatchNorm Mixed_6a_Branch_1_Conv2d_1a_3x3_BatchNorm Block17_1_Branch_1_Conv2d_0a_1x1 Block17_1_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_1_Branch_1_Conv2d_0b_1x7 Block17_1_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_1_Branch_0_Conv2d_1x1 Block17_1_Branch_1_Conv2d_0c_7x1 Block17_1_Branch_0_Conv2d_1x1_BatchNorm Block17_1_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_1_Conv2d_1x1 Block17_2_Branch_1_Conv2d_0a_1x1 Block17_2_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_2_Branch_1_Conv2d_0b_1x7 Block17_2_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_2_Branch_0_Conv2d_1x1 Block17_2_Branch_1_Conv2d_0c_7x1 Block17_2_Branch_0_Conv2d_1x1_BatchNorm Block17_2_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_2_Conv2d_1x1 Block17_3_Branch_1_Conv2d_0a_1x1 Block17_3_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_3_Branch_1_Conv2d_0b_1x7 Block17_3_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_3_Branch_0_Conv2d_1x1 Block17_3_Branch_1_Conv2d_0c_7x1 Block17_3_Branch_0_Conv2d_1x1_BatchNorm Block17_3_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_3_Conv2d_1x1 Block17_4_Branch_1_Conv2d_0a_1x1 Block17_4_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_4_Branch_1_Conv2d_0b_1x7 Block17_4_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_4_Branch_0_Conv2d_1x1 Block17_4_Branch_1_Conv2d_0c_7x1 Block17_4_Branch_0_Conv2d_1x1_BatchNorm Block17_4_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_4_Conv2d_1x1 Block17_5_Branch_1_Conv2d_0a_1x1 Block17_5_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_5_Branch_1_Conv2d_0b_1x7 Block17_5_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_5_Branch_0_Conv2d_1x1 Block17_5_Branch_1_Conv2d_0c_7x1 Block17_5_Branch_0_Conv2d_1x1_BatchNorm Block17_5_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_5_Conv2d_1x1 Block17_6_Branch_1_Conv2d_0a_1x1 Block17_6_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_6_Branch_1_Conv2d_0b_1x7 Block17_6_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_6_Branch_0_Conv2d_1x1 Block17_6_Branch_1_Conv2d_0c_7x1 Block17_6_Branch_0_Conv2d_1x1_BatchNorm Block17_6_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_6_Conv2d_1x1 Block17_7_Branch_1_Conv2d_0a_1x1 Block17_7_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_7_Branch_1_Conv2d_0b_1x7 Block17_7_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_7_Branch_0_Conv2d_1x1 Block17_7_Branch_1_Conv2d_0c_7x1 Block17_7_Branch_0_Conv2d_1x1_BatchNorm Block17_7_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_7_Conv2d_1x1 Block17_8_Branch_1_Conv2d_0a_1x1 Block17_8_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_8_Branch_1_Conv2d_0b_1x7 Block17_8_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_8_Branch_0_Conv2d_1x1 Block17_8_Branch_1_Conv2d_0c_7x1 Block17_8_Branch_0_Conv2d_1x1_BatchNorm Block17_8_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_8_Conv2d_1x1 Block17_9_Branch_1_Conv2d_0a_1x1 Block17_9_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_9_Branch_1_Conv2d_0b_1x7 Block17_9_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_9_Branch_0_Conv2d_1x1 Block17_9_Branch_1_Conv2d_0c_7x1 Block17_9_Branch_0_Conv2d_1x1_BatchNorm Block17_9_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_9_Conv2d_1x1 Block17_10_Branch_1_Conv2d_0a_1x1 Block17_10_Branch_1_Conv2d_0a_1x1_BatchNorm Block17_10_Branch_1_Conv2d_0b_1x7 Block17_10_Branch_1_Conv2d_0b_1x7_BatchNorm Block17_10_Branch_0_Conv2d_1x1 Block17_10_Branch_1_Conv2d_0c_7x1 Block17_10_Branch_0_Conv2d_1x1_BatchNorm Block17_10_Branch_1_Conv2d_0c_7x1_BatchNorm Block17_10_Conv2d_1x1 Mixed_7a_Branch_2_Conv2d_0a_1x1 Mixed_7a_Branch_2_Conv2d_0a_1x1_BatchNorm Mixed_7a_Branch_0_Conv2d_0a_1x1 Mixed_7a_Branch_1_Conv2d_0a_1x1 Mixed_7a_Branch_2_Conv2d_0b_3x3 Mixed_7a_Branch_0_Conv2d_0a_1x1_BatchNorm Mixed_7a_Branch_1_Conv2d_0a_1x1_BatchNorm Mixed_7a_Branch_2_Conv2d_0b_3x3_BatchNorm Mixed_7a_Branch_0_Conv2d_1a_3x3 Mixed_7a_Branch_1_Conv2d_1a_3x3 Mixed_7a_Branch_2_Conv2d_1a_3x3 Mixed_7a_Branch_0_Conv2d_1a_3x3_BatchNorm Mixed_7a_Branch_1_Conv2d_1a_3x3_BatchNorm Mixed_7a_Branch_2_Conv2d_1a_3x3_BatchNorm Block8_1_Branch_1_Conv2d_0a_1x1 Block8_1_Branch_1_Conv2d_0a_1x1_BatchNorm Block8_1_Branch_1_Conv2d_0b_1x3 Block8_1_Branch_1_Conv2d_0b_1x3_BatchNorm Block8_1_Branch_0_Conv2d_1x1 Block8_1_Branch_1_Conv2d_0c_3x1 Block8_1_Branch_0_Conv2d_1x1_BatchNorm Block8_1_Branch_1_Conv2d_0c_3x1_BatchNorm Block8_1_Conv2d_1x1 Block8_2_Branch_1_Conv2d_0a_1x1 Block8_2_Branch_1_Conv2d_0a_1x1_BatchNorm Block8_2_Branch_1_Conv2d_0b_1x3 Block8_2_Branch_1_Conv2d_0b_1x3_BatchNorm Block8_2_Branch_0_Conv2d_1x1 Block8_2_Branch_1_Conv2d_0c_3x1 Block8_2_Branch_0_Conv2d_1x1_BatchNorm Block8_2_Branch_1_Conv2d_0c_3x1_BatchNorm Block8_2_Conv2d_1x1 Block8_3_Branch_1_Conv2d_0a_1x1 Block8_3_Branch_1_Conv2d_0a_1x1_BatchNorm Block8_3_Branch_1_Conv2d_0b_1x3 Block8_3_Branch_1_Conv2d_0b_1x3_BatchNorm Block8_3_Branch_0_Conv2d_1x1 Block8_3_Branch_1_Conv2d_0c_3x1 Block8_3_Branch_0_Conv2d_1x1_BatchNorm Block8_3_Branch_1_Conv2d_0c_3x1_BatchNorm Block8_3_Conv2d_1x1 Block8_4_Branch_1_Conv2d_0a_1x1 Block8_4_Branch_1_Conv2d_0a_1x1_BatchNorm Block8_4_Branch_1_Conv2d_0b_1x3 Block8_4_Branch_1_Conv2d_0b_1x3_BatchNorm Block8_4_Branch_0_Conv2d_1x1 Block8_4_Branch_1_Conv2d_0c_3x1 Block8_4_Branch_0_Conv2d_1x1_BatchNorm Block8_4_Branch_1_Conv2d_0c_3x1_BatchNorm Block8_4_Conv2d_1x1 Block8_5_Branch_1_Conv2d_0a_1x1 Block8_5_Branch_1_Conv2d_0a_1x1_BatchNorm Block8_5_Branch_1_Conv2d_0b_1x3 Block8_5_Branch_1_Conv2d_0b_1x3_BatchNorm Block8_5_Branch_0_Conv2d_1x1 Block8_5_Branch_1_Conv2d_0c_3x1 Block8_5_Branch_0_Conv2d_1x1_BatchNorm Block8_5_Branch_1_Conv2d_0c_3x1_BatchNorm Block8_5_Conv2d_1x1 Block8_6_Branch_1_Conv2d_0a_1x1 Block8_6_Branch_1_Conv2d_0a_1x1_BatchNorm Block8_6_Branch_1_Conv2d_0b_1x3 Block8_6_Branch_1_Conv2d_0b_1x3_BatchNorm Block8_6_Branch_0_Conv2d_1x1 Block8_6_Branch_1_Conv2d_0c_3x1 Block8_6_Branch_0_Conv2d_1x1_BatchNorm Block8_6_Branch_1_Conv2d_0c_3x1_BatchNorm Block8_6_Conv2d_1x1 Bottleneck

ValueError Traceback (most recent call last) in 9 weights.append(weight_arr) 10 print(layer.name) ---> 11 layer.set_weights(weights) 12 13 print('Saving weights...')

D:\Anaconda3\lib\site-packages\keras\engine\base_layer.py in set_weights(self, weights) 1055 str(pv.shape) + 1056 ' not compatible with ' -> 1057 'provided weight shape ' + str(w.shape)) 1058 weight_value_tuples.append((p, w)) 1059 K.batch_set_value(weight_value_tuples)

ValueError: Layer weight shape (1792, 128) not compatible with provided weight shape (1792, 512)`

[stezarpriansya]

i face the same problem too, anyone can help?

[jyun-bunny-honey]

model = InceptionResNetV1(classes=512) solved the problem. My guess is that newly updated model was not trained under the default frame of Inception ResNet V1.

[stezarpriansya]

model = InceptionResNetV1(classes=512) solved the problem. My guess is that newly updated model was not trained under the default frame of Inception ResNet V1.

thank you 👍

[jyun-bunny-honey]

model = InceptionResNetV1(classes=512) solved the problem. My guess is that newly updated model was not trained under the default frame of Inception ResNet V1.

I found the answer from the link: https://jekel.me/2018/512_vs_128_facenet_embedding_application_in_Tinder_data/. I rechecked the model structure, and the facenet model I used (20180408-102900) does have 512 embeddings. I think possibly some codes in demos need to be updated accordingly as well.