ValueError: Scorer initialization failed with error code 8198 swig/python detected a memory leak of type 'Alphabet *', no destructor found

DeepSpeech
#21

As I said, it’s not yet ready, I’m facing the same issue but my time is limited, was busy with other stuff, now it’s weekend.

#22

Maybe it’s caused by this manual virtual env activation? Try prepending the Python commands with a call to the activation script, like so:

RUN python3 -m venv venv/
RUN source venv/bin/activate && pip install --upgrade pip==20.0.2 wheel==0.34.2 setuptools==46.1.3
RUN source venv/bin/activate && pip install --upgrade --editable .
RUN source venv/bin/activate && python DeepSpeech.py ...
#23

No problem. I just wanted to raise the issue now and see if it was a similar issue you were having. I’ll give Reuben’s suggestion a try in the meantime.

#24

I don’t think source works well in Dockerfile. I think you have to use the dot operator like:

RUN python3 -m venv venv/
RUN . venv/bin/activate && pip install --upgrade pip==20.0.2 wheel==0.34.2 setuptools==46.1.3
RUN . venv/bin/activate && pip install --upgrade --editable .
RUN . venv/bin/activate && python DeepSpeech.py ...

I get “bin/sh/ source” not found errors every time I’ve tried using source to activate a venv in Docker. I’m sure there are other methods using SHELL but not sure…

#25

So I had successful build and training by doing something like this. The next step would be to activate the venv in /DeepSpeech/venv/bin/activate and run the training .sh script for full automation but with this Dockerfile you’ll get all the desired venv scripts in your /DeepSpeech directory and training will be ready.

Thanks for your help @reuben and @lissyx . Appreciate it.

FROM tensorflow/tensorflow:1.15.2-py3

RUN apt-get update && apt-get install -y \
        apt-utils \
        vim \
        bash-completion \
        build-essential \
        curl \
        git \
        git-lfs \
        unzip \
        wget \
        python3-venv

# Change to Root and install git lfs alternative way
WORKDIR /
RUN git lfs install
RUN git clone  https://github.com/mozilla/DeepSpeech

#  Change to Deepspeech repo clone and setup venv and pip3 install all of the training code
WORKDIR /DeepSpeech
#RUN python3 -m venv venv/
#ENV VIRTUAL_ENV $(pwd)/venv
#ENV PATH $VIRTUAL_ENV/bin:$PATH
#RUN which python3
RUN python3 -m venv venv/
RUN . venv/bin/activate && pip install --upgrade pip==20.0.2 wheel==0.34.2 setuptools==46.1.3
RUN . venv/bin/activate && pip install --upgrade --editable .
RUN which python3
#26

I am getting the similar scorer error and when running git lfs pull, I get the following error:

batch response: This repository is over its data quota. Account responsible for LFS bandwidth should purchase more data packs to restore access.
error: failed to fetch some objects from 'https://github.com/mozilla/DeepSpeech.git/info/lfs'

I am working with DeepSpeech-0.7.4 on Colab

#27

We have disabled Git LFS, please stop using it.

1 Like
#28

I have simply cloned the repo but it still gives
ValueError: Scorer initialization failed with error code 8198
Although the kenlm.scorer is present in the default directory, would be really thankful if you can help.

#29

Sorry but you hijack old threads without sharing any informations. As I said, we stopped using Git LFS, please uninstall it, and download kenlm.scorer from 0.7.4 github release artifacts

#30

I apologize, I am new here, saw the same issue and thought this was the right place to post. The issue has been resolved by manually downloading the scorer. thanks!

1 Like