No module named 'deepspeech_training'

Hi All,
While running deepspeech.py I am getting this No module named ‘deepspeech_training’ error.
Recently the path to deepspeech_training has been changed to training/deepspeech_training so,I think we have to make change in the deepspeech.py file also??

from training.deepspeech_training import train as ds_train

1 Like

Python imports will kill me someday :slight_smile:

Please look at how you are calling the DeepSpeech.py
script. If you do that in deepspeech_training you should be fine. But I guess you are calling it from somewhere else?

If you need more help, post more info.

Have you read the documentation and performed pip install -e . step ?

1 Like

No I have not performed this step pip install -e …In the documentation “using pre-trained model” steps I have performed.What is "pip install -e "?

This is the section about using deepspeech for running inference. Please read the training documentation if you are intending to perform training?

Hi, facing the same issue.
I followed the training doc and redid the installation multiple times and still facing the same issue.
Please guide me through the ‘pip install -e’ step, you have mentioned because i couldn’t find it in the training docs.

here is the error, i received. I am doing the training on google colab, so I haven’t created a virtual env.

/content/drive/My Drive/transfer learning/deepspeech
Training package is not installed. See training documentation.
Traceback (most recent call last):
File “DeepSpeech.py”, line 7, in
from deepspeech_training import train as ds_train
ModuleNotFoundError: No module named ‘deepspeech_training’

It is right here: https://deepspeech.readthedocs.io/en/v0.7.0/TRAINING.html#installing-deepspeech-training-code-and-its-dependencies

We can’t help for Google Colab.

hi,
Thanks for the prompt reply.
The issue has been solved from the doc that you mentioned. Sorry, for the inconvenience.

There was a mismatch between the version i was using and the document i was referring to.

1 Like

Please send PR if you feel doc improvement is needed :slight_smile:

Is there a conda equivalence for pip3 install --upgrade --force-reinstall -e .? Thank you!

@Vibhav can you please share that google colab link where you run deepspeech model.

here you go
https://colab.research.google.com/drive/19-7L0hfK_KfeRstG5f2MqAcqfAQTt5Jc#scrollTo=YhFyawDY4qw8

Ran into this issue trying today, with the latest git version, also tried with a checkout of tags/v0.7.4 with the same result.

i’m at the top level of the project and calling ./bin/run-ldc93s1.sh and it says it cannot find deepspeech_training.

Have created a venv of python 3.6, activated the venv and have installed a new pip, wheel and setup-tools, as well as tensorflow-gpu and deepspeech-gpu.

Please dont, as documented this is the GPU backed inference python module. Please follow the docs for installing the training code: https://deepspeech.readthedocs.io/en/v0.7.4/TRAINING.html#installing-deepspeech-training-code-and-its-dependencies

Thanks, that’s exactly what I followed. i didn’t install deepspeech-gpu initially, but once it failed i tried it just in case it helped (which it didn’t). I’m now seeing that even running ./DeepSpeech gives the same error, at the top level.

Training package is not installed. See training documentation.
Traceback (most recent call last):
  File "./DeepSpeech.py", line 7, in <module>
    from deepspeech_training import train as ds_train
ModuleNotFoundError: No module named 'deepspeech_training'

Well, you did not mentionned it in your previous message:

Have created a venv of python 3.6, activated the venv and have installed a new pip, wheel and setup-tools, as well as tensorflow-gpu and deepspeech-gpu.

Are you sure that pip3 install --upgrade -e . completed successfully? Are you sure there is no mixup of python version here?

You’re right, I didn’t use the -e step, it pulled in a bunch of new dependencies, that’s interesting! I was looking for a requirements file to see how to pull in requirements but i couldn’t find it, and I see that -e must be doing something like that. my mistake.

i’ve got another error now,
> File “/home/foo/Projects/DeepSpeech/venv/lib/python3.6/site-packages/ds_ctcdecoder/init.py”, line 42, in init
> raise ValueError(‘Scorer initialization failed with error code {}’.format(err))
> ValueError: Scorer initialization failed with error code 8198

Yes, we moved away from requirements.txt, dependencies are expressed in setup.py because it allows more flexibility as well.

Often a case of improper checkout because of missing git-lfs

That would be it, i’ll investigate the LFS. thank you so much.

8198 == 0x2006:

APPLY(DS_ERR_SCORER_INVALID_LM, 0x2006, “Could not recognize language model header in scorer.”)
https://deepspeech.readthedocs.io/en/v0.7.3/Error-Codes.html