Investigate using LLMs for evaluation

[eu9ene]

It would be interesting to compare evaluation capabilities of LLMs to COMET and human evaluation.

See the paper: Large Language Models Are State-of-the-Art Evaluators of Translation Quality