Setup environment using Conda

1. Initialize a new conda environment (e.g., using Python 3.9).

2. Activate the conda environment by executing the command in your terminal:

   $ conda activate <conda-env-name>

3. Install pytorch by executing the command below:

$ python -m pip install torch==1.8.2+cu111 torchvision==0.9.2+cu111 torchaudio==0.8.2 -f https://download.pytorch.org/whl/lts/1.8/torch_lts.html

3. Install allennlp and HuggingFace datasets using the requirements.txt, by executing the following command:

$ python -m pip install -r requirements.txt

Note: In the outline above, we assume you are have CULDA TOOLKIT 11. You will have to update the version in step 2 accordingly, in case your version differs. Moreover, note that since allennlp version 2.5.0 requires pytorch 1.8, we opt for pre-installing that version in our environment.

Running

$ allennlp train training_config/qasper_eqqa.jsonnet -s data/qasper_models --include-package quality_estimation