huggingface / huggingface/evaluate
Add common metrics for information retrieval
Open
metric request
- Dominant language
- Python
- Stars
- 2.5k
- Forks
- 341
- PR merge metrics
- No merged PRs in 30d
Description
There's a nice summary of common metrics for IR here: https://www.pinecone.io/learn/offline-evaluation/
Although we have many of these as part of `trec_eval`, it could make sense to have separate metrics like MRR and NDCG@K to give visibility
Contributor guide
Assessment
This issue has not been assessed yet.