Training on Urdu giving eager execution error

DeepSpeech
#1

Hello,

I am working on creating a speech to text engine for Urdu (Pakistani national language). I am using DeepSpeech 0.9.1 and followed the instructions in the documentation. The execution environment is Colab.

When running the command below, I got an error that I’ve not been able to resolve.

! python3 DeepSpeech.py \
        --drop_source_layers 1 \
        --n_hidden 2048 \
        --alphabet_config_path /content/gdrive/My\ Drive/dataset/UrduAlphabet_newscrawl2.txt \
        --checkpoint_dir /content/gdrive/My\ Drive/dataset/trained_load_checkpoint \
        --train_files  /content/gdrive/My\ Drive/train.csv \
        --dev_files   /content/gdrive/My\ Drive/dev.csv \
        --test_files  /content/gdrive/My\ Drive/test.csv \
        --epochs 10 \
        --train_batch_size 32 \
        --export_dir /content/gdrive/MyDrive/dataset/urdu_trained \
        --export_file_name urdu \
        --test_batch_size 12 \
        --learning_rate 0.00001 \
        --load_cudnn true \
        --reduce_lr_on_plateau true \
        --scorer /content/gdrive/My\ Drive/urdu_lm/kenlm.scorer

The error I am getting is:

I1126 22:10:59.237078 139950181578624 utils.py:141] NumExpr defaulting to 2 threads.
    Traceback (most recent call last):
      File "DeepSpeech.py", line 12, in <module>
        ds_train.run_script()
      File "/content/DeepSpeech/training/deepspeech_training/train.py", line 976, in run_script
        absl.app.run(main)
      File "/usr/local/lib/python3.6/dist-packages/absl/app.py", line 300, in run
        _run_main(main, args)
      File "/usr/local/lib/python3.6/dist-packages/absl/app.py", line 251, in _run_main
        sys.exit(main(argv))
      File "/content/DeepSpeech/training/deepspeech_training/train.py", line 948, in main
        train()
      File "/content/DeepSpeech/training/deepspeech_training/train.py", line 460, in train
        dropout_rates = [tfv1.placeholder(tf.float32, name='dropout_{}'.format(i)) for i in range(6)]
      File "/content/DeepSpeech/training/deepspeech_training/train.py", line 460, in <listcomp>
        dropout_rates = [tfv1.placeholder(tf.float32, name='dropout_{}'.format(i)) for i in range(6)]
      File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/array_ops.py", line 2616, in placeholder
        raise RuntimeError("tf.placeholder() is not compatible with "
    RuntimeError: tf.placeholder() is not compatible with eager execution.

I have tried disabling eager execution but it has no effect on this package.

Thank you for your help.

#2

As you don’t share anything about your environment it is hard to tell, look at the guidelines and post accordingly if you want us to help.

#3

You have not setup a virtualenv as documented.

We don’t use that, we can’t provide support.

#4

Hello saima!

i had the exact error and it had been caused by not Restarting the runtime after i updated and installed the latest tensorflow-gpu : ‘tensorflow-gpu==1.15.4’ and also not Restarting the runtime after :

!pip3 install --upgrade pip==20.2.2 wheel==0.34.2 setuptools==49.6.0

!pip3 install --upgrade -e .

it started the training after i restarted my runtime after these.