Help contribute to dl-translate

[xhluca]

If you enjoyed using dl-translate and want to make it better, feel free to reach out to contribute to dl-translate! Although I have initially developed the library, I believe that others have certain skills to help make this library better. I'd be happy to add a "contributors" section somewhere in the readme describing what everyone have done once we have a certain number of contributors.

To get started, check out the issues with the "help wanted" tag.

Additionally, there's other areas to contribute:

• The documentations can be improved. for example, readthedocs is pretty finicky because of recent dependency issues. I'm also using jinja templating but a better system could be adopted (e.g. pandas docs).
• More test cases would be welcome for scenarios that are not currently covered.
• Adding recent models like NLLB-200 by FAIR
• A user interface for local translation
• More efficient optimization (e.g. ONNX conversion) for real-time translation on lower-end hardware
• Using techniques like distillation to speed up inference of base models without losing out on accuracy