Building word2vec

Building word2vec

The instructions provided below specify the steps to build word2vec version 0.1c on Linux on IBM Z for following distributions:

• RHEL (7.8, 7.9, 8.4, 8.6, 9.0)
• SLES (12 SP5, 15 SP3, 15 SP4)
• Ubuntu (18.04, 20.04, 22.04)

General notes:

• When following the steps below please use a standard permission user unless otherwise specified.

• A directory /<source_root>/ will be referred to in these instructions, this is a temporary writable directory anywhere you'd like to place it.

Build word2vec

1. Install standard utilities, packages and platform specific dependencies

• RHEL (7.8, 7.9, 8.4, 8.6, 9.0)

   sudo yum install -y gcc make wget tar unzip

• SLES (12 SP5, 15 SP3, 15 SP4)

   sudo zypper install -y gcc make wget tar unzip

• Ubuntu (18.04, 20.04, 22.04)

   sudo apt-get update
   sudo apt-get install -y gcc make wget tar unzip

2. Create a working directory and download word2vec source code

    cd $SOURCE_ROOT
    wget https://storage.googleapis.com/google-code-archive-source/v2/code.google.com/word2vec/source-archive.zip
    unzip source-archive.zip
   

3. Build word2vec

    cd word2vec/trunk
    make CFLAGS="-lm -pthread -O3 -Wall -funroll-loops"
   

4. Set environment variables

    export PATH=$PATH:$SOURCE_ROOT/word2vec/trunk
   

5. Test word2vec using demo scripts

    ./demo-word.sh
    ./demo-phrases.sh
   

   Note: Enter test corpus as input and get word vectors as output, e.g. Input=france

6. Run word2vec binary

    word2vec
   

   Note: The word2vec tool takes a text corpus as input and produces the word vectors as output.

References:

• https://code.google.com/archive/p/word2vec/