An example of a machine translation model that translates Dyula to French by finetuning a pre-trained the t5-small
model.
Make sure you run your notebooks in the relevant virtual environments created below.
Set up your environment with the required dependencies.
-
For data processing and training the model, set up the
train
environment by running the following:# Use Python 3.10 # Initial setup python -m venv train-venv && source train-venv/bin/activate pip install -r notebooks/requirements.txt # Activate after setup (run every time) source train-venv/bin/activate
-
For model serving and inference, set up the
serve
environment by running the following:Make sure you uncomment the
ipykernel
requirement in therequirements.txt
file before running the commands below if you want to run the inference notebook.# Use Python 3.10 # Initial setup python -m venv serve-venv && source serve-venv/bin/activate pip install -r deployment/requirements.txt # Activate after setup (run every time) source serve-venv/bin/activate