OpenGPT-X with BigScience

This repository contains set-up code for working on JSC machines with the BigScience codebase.

Aims

We aim to support a set-up for your own, specialized working environment. This means we do not fix versions or commits; this would have to be done on an individual/subgroup level.

Getting Started

Setting Up

We assume you have already set up your environment on the supercomputer. If you have not, please see the "Getting started at JSC" guide. guide.

First off, clone this repository to a destination of your choice. Adjust variables in ./run_scripts/variables.bash and execute:

cd run_scripts
nice bash set_up.bash

For weird reasons, sometimes Bash parses incorrectly and the script breaks. Please just try again in that case.

Starting Training

In ./run_scripts/tr1-13B-round1_juwels_pipe.sbatch, adjust the #SBATCH variables on top as desired (most interesting is the number of --nodes) and execute:

cd run_scripts
sbatch tr1-13B-round1_juwels_pipe.sbatch

Please always run the scripts from the run_scripts directory. We have not yet made them execution-location-independent.

Care needs to be taken when changing the number of GPUs per node. You also need to change the GPUS_PER_NODE variable accordingly, as we do not yet bother with parsing the SLURM_GRES value.

The script we currently work with, ./run_scripts/tr1-13B-round1_juwels_pipe.sbatch, is the oldest training sbatch script from the BigScience documentation repository. This matches the current data structure we use for testing; a newer version that assumes later PyTorch versions has different data structure requirements due to the many different corpora BigScience is training on.

PyTorch >= 1.11 will complain about not being able to handle some address families and tell you that sockets are invalid. This does not hinder the code from scaling according to the number of total GPUs.

Supported Machines

• JUWELS Cluster
• JUWELS Booster

Supported means tested and the correct CUDA compute architecture will be selected. Other machines can easily be supported by adjusting ./run_scripts/activate.bash.