Explain metrics from `experiments/eval.py` in `README.md`

It could be nice to have in the `README.md` a section explaining the `eval.py` file and the different evaluation metrics.

Start by reading experiments/eval.py to identify the evaluation metrics and then inspect README.md for the appropriate section and existing terminology. Document what eval.py does and explain each metric in README.md; the work is complete when a reader can understand the file and its metrics from the README alone.

Written by the indexing model from the issue text.