TypeError: Cannot handle this data type

[OswaldoBornemann]

when i ran python train.py --output_directory=taco_output/ --log_directory=log_dir
it shown me the below error, seems the tensorboardX not work.

FP16 Run: False
Dynamic Loss Scaling: True
Distributed Run: False
cuDNN Enabled: True
cuDNN Benchmark: False
Epoch: 0
Train loss 0 76.131721 Grad Norm 14.724035 3.98s/it
Validation loss 0: 65.199608
Traceback (most recent call last):
  File "/data/anaconda3/lib/python3.6/site-packages/PIL/Image.py", line 2428, in fromarray
    mode, rawmode = _fromarray_typemap[typekey]
KeyError: ((1, 1, 400), '|u1')

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "train.py", line 284, in <module>
    args.warm_start, args.n_gpus, args.rank, args.group_name, hparams)
  File "train.py", line 242, in train
    hparams.distributed_run, rank)
  File "train.py", line 147, in validate
    logger.log_validation(reduced_val_loss, model, y, y_pred, iteration)
  File "/data/sh_tacotron2/logger.py", line 34, in log_validation
    iteration)
  File "/data/anaconda3/lib/python3.6/site-packages/tensorboardX/writer.py", line 427, in add_image
    image(tag, img_tensor, dataformats=dataformats), global_step, walltime)
  File "/data/anaconda3/lib/python3.6/site-packages/tensorboardX/summary.py", line 216, in image
    image = make_image(tensor, rescale=rescale)
  File "/data/anaconda3/lib/python3.6/site-packages/tensorboardX/summary.py", line 254, in make_image
    image = Image.fromarray(tensor)
  File "/data/anaconda3/lib/python3.6/site-packages/PIL/Image.py", line 2431, in fromarray
    raise TypeError("Cannot handle this data type")
TypeError: Cannot handle this data type

[OswaldoBornemann]

i downgrade the tensorboardX to 1.1 and then work.

[delgerdalai]

pip install -r requirements.txt

[batho2n]

But, If tensorboardX==1.1, following error message is occured

TypeError: Conflict register for file "tensorboard/compat/proto/resource_handle.proto": 
tensorboard.ResourceHandleProto is already defined in file "tensorboardX/src/resource_handle.proto". 
Please fix the conflict by adding package name on the proto file, or use different name for the 
duplication.

[nimamoradi]

i have same exact issue

[davislf2]

@batho2n did you resolve it? I have the same error of "Conflict register" in tensorboard and tensorboardX as well.

[batho2n]

@davislf2, I resolved the problem by removing tensorboard package

$ pip uninstall tensorboard

[bfs18]

add dataformats='HWC' argument to Tacotron2Logger self.add_image function. The default value is 'CHW', which is not consistent with the image generated by plot_alignment_to_numpy.