Tacotron-2: error in loading checkpoints

Created on 3 Aug 2018  路  2Comments  路  Source: Rayhane-mamah/Tacotron-2

I have download the checkpoint file and used it in synthesis, and I get this error:

NotFoundError (see above for traceback): Key model/inference/decoder/Location_Sensitive_Attention/location_features_convolution/bias not found in checkpoint
[[Node: save/RestoreV2_2 = RestoreV2[dtypes=[DT_FLOAT], _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_save/Const_0_0, save/RestoreV2_2/tensor_names, save/RestoreV2_2/shape_and_slices)]]
[[Node: save/RestoreV2_27/_145 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/device:GPU:0", send_device="/job:localhost/replica:0/task:0/device:CPU:0", send_device_incarnation=1, tensor_name="edge_296_save/RestoreV2_27", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/device:GPU:0"]()]]

Most helpful comment

Hello @wanglu0609 thanks for reaching out.

I am assuming you used my pretrained model I used few versions earlier. Please keep in mind that the repo changed since that model and the model architecture with it so the saved graph and the one constructed are inconsistent. I am doing a fresh training on LJspeech, I will upload it once it's done.

I will notify you once the pretrained model is online. :)

All 2 comments

Hello @wanglu0609 thanks for reaching out.

I am assuming you used my pretrained model I used few versions earlier. Please keep in mind that the repo changed since that model and the model architecture with it so the saved graph and the one constructed are inconsistent. I am doing a fresh training on LJspeech, I will upload it once it's done.

I will notify you once the pretrained model is online. :)

@Rayhane-mamah , any news on that new pre-trained model? Or could you tell us which commit to go back to get this dataset to run

Was this page helpful?
0 / 5 - 0 ratings