readme

README.md

@@ -7,6 +7,8 @@ Documentation is work in progress and it will improve over the following months,
 
 Model description and evaluation results are also in preparation.
 
+In order to run PIE, the easiest way is to download the repository and install the dependencies (see requirements.txt). Training models is done with `python train.py path/to/config.json`. All non-nested parameters can be overwritten directly from the command line using environment variables like `PIE_DEVICE=cpu` (for input parameter `device`. Warning: bear in mind that `PIE_...=False` will be parsed into a boolean `True`, in order to get `False` you could use `PIE_...=""`).
+
 - Future work:
   - Add GRL regularization on domain/source labels (which seems to help POS [https://arxiv.org/pdf/1805.06093.pdf](here: Table 1&2). We can use file names, or derive appropriate labels from file names.
   - Implement a mixture-model to learn to decide whether to retrieve a lemma from the cache or go and generate it one character at a time [https://arxiv.org/pdf/1609.07843.pdf](similar to this paper).