brq-att-alt-exp

repository for experimenting with alternative forms of attention and best-rq

instructions

• create env with speechbrain (see below)
• set up yaml file
  • for examples of pretraining see haparams folder
  • for examples of finetuning see the finetune
• there are two files that are needed for running the model in speechbrain
  • mask.py (used in pretraining see train.py)
  • quantiser.py (used in yaml file)

conda create --name NAME_OF_ENV python=3.11
conda activate NAME_OF_ENV

pip install -r requirements.txt
pip install -e .