haiku.rag/evaluations/tests
Yiorgis Gozadinos 2bbd949a29
Record the database each citation came from
A run over several databases could report which documents were cited but
not which database grounded the answer: `_result_from_run` walked the
citation index for `document_uri` and dropped `Citation.source`. The
distribution is not recoverable from the report afterwards, so a sharded
run would have measured everything except attribution.

`cited_sources` is one entry per cited chunk, in citation order, empty
where the database is unnamed.
2026-08-24 10:03:47 +03:00
..
__init__.py Tests for gepa 2026-03-12 12:05:25 +02:00
test_benchmark.py Split the evaluation benchmark by responsibility 2026-08-20 14:08:09 +03:00
test_capability_runner.py Record the database each citation came from 2026-08-24 10:03:47 +03:00
test_citation_evaluators.py Add MTRAG ClapNQ multi-turn evaluation 2026-08-17 10:53:16 +03:00
test_config.py Evaluate over a configured set of databases 2026-08-24 10:03:47 +03:00
test_conversation_evaluator.py Add MTRAG ClapNQ multi-turn evaluation 2026-08-17 10:53:16 +03:00
test_datasets.py Add stable ids to FRAMES question rows 2026-08-24 09:03:44 +03:00
test_evaluators.py Add MTRAG ClapNQ multi-turn evaluation 2026-08-17 10:53:16 +03:00
test_mtrag.py Simplify the eval harness and share the embed-fill path. 2026-08-17 10:56:26 +03:00
test_numbers.py Normalize unicode signs 2026-06-06 14:52:04 +03:00
test_reference_configs.py Pin the eval judge to qwen3.8 2026-08-18 14:28:16 +03:00
test_submission.py Add T²-RAGBench leaderboard submission exporter 2026-06-08 11:51:02 +03:00