- “mySoftware” represents your own code or python script you want to run. You need to replace “mySoftware” with the name of your script.
- “$inputDataset” defines the path to the input document pairs. You can choose three datasets: small (dataset 2), large (dataset 1) for debugging, or test set.
- $outputDir takes the path where to store your anwers.jsonl file.
In my case, I only need to define the absolute path of my python script. Alternatively, you can name the working directory. Here is an example of my command line, how it works for me (using the absolute path):
/home/boenninghoff21/miniconda3/envs/pan/bin/python /home/boenninghoff21/final/main_inference.py -i $inputDataset -o $outputDir
As you can see, I installed a python environment via miniconda. And my script “main_inference.py” takes two arguments for input data and output files. I would suggest you to test it with dataset 1 or 2. Then you will get feedback if any error occurs. With the test data, everything is hidden.
Hope, it was helpful.