pytorch-facenet

train.py

C:\Users\liu\Anaconda3\envs\pytorch01\python.exe F:/facenet/facenet-pytorch/facenet-pytorch/train.py
WARNING:root:This caffe2 python run does not have GPU support. Will run in CPU only mode.
Load weights model_data/facenet_mobilenet.pth.

Successful Load Key: ['backbone.model.stage1.0.0.weight', 'backbone.model.stage1.0.1.weight', 'backbone.model.stage1.0.1.bias', 'backbone.model.stage1.0.1.running_mean', 'backbone.model.stage1.0.1.running_var', 'backbone.model.stage1.0.1.num_batches_tracked', 'backbone.model.stage1.1.0.weight', 'backbone.model.stage1.1.1.weight', 'backbone.model.stage1.1.1.bias', 'backbone.model.stage1.1.1.running_mean', 'backbone.model.stage1.1.1.running_var', 'backbone.model.stage1.1.1.num_batches_tracked', 'backbone.model.stage1. ……
Successful Load Key Num: 168

Fail To Load Key: ['classifier.weight', 'classifier.bias'] ……
Fail To Load Key num: 2
F:\facenet\facenet-pytorch\facenet-pytorch\utils\dataloader.py:164: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return np.array(pairs)
Configurations:
----------------------------------------------------------------------
|                     keys |                                   values|
----------------------------------------------------------------------
|              num_classes |                                     4999|
|                 backbone |                                mobilenet|
|               model_path |         model_data/facenet_mobilenet.pth|
|              input_shape |                            [160, 160, 3]|
|               Init_Epoch |                                        0|
|                    Epoch |                                       50|
|               batch_size |                                        6|
|                  Init_lr |                                    0.001|
|                   Min_lr |                                    1e-05|
|           optimizer_type |                                     adam|
|                 momentum |                                      0.9|
|            lr_decay_type |                                      cos|
|              save_period |                                        1|
|                 save_dir |                                     logs|
|              num_workers |                                        0|
|                num_train |                                   256594|
|                  num_val |                                     2591|
----------------------------------------------------------------------
Start Train
Epoch 1/50:   0%|          | 55/42765 [15:32<202:08:13, 17.04s/it, accuracy=0, lr=3e-5, total_CE_loss=11.8, total_triple_loss=0