danvics/haiku.rag
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fold the changelog sections back together

Browse source 
The rebase onto lancedb 0.37.1 left duplicate Added and Fixed headers where
both sides had entries. Same 14 entries, four sections.

Claude-Session: https://claude.ai/code/session_01WhudUtZm6qqiuv8Y1sbwSc
This commit is contained in:
Yiorgis Gozadinos 2026-08-31 19:10:10 +03:00
parent 058120c05a
commit 33bd0be702
No known key found for this signature in database
1 changed files with 4 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
CHANGELOG.md
View file

@ -13,24 +13,19 @@
text untruncated. `format_citations_rich` takes a `full` argument.
- `doctor` fails when the chunks FTS index covers no rows.
- FTS and hybrid searches log a warning when the FTS index covers no rows.
- `evaluations run --retrieval-limit N`: candidates each database fetches during the retrieval benchmark, overriding the dataset's `retrieval_limit`.
- `mtrag_pooled` evaluation dataset and its reference config `evaluations/configs/mtrag_pooled.yaml`: all four MTRAG domains pooled and partitioned across `n` collections, `--alpha` interpolating between one domain per collection and a uniform shard.
- `mtrag_federated` evaluation dataset and its reference config `evaluations/configs/mtrag_federated.yaml`: MTRAG ClapNQ partitioned by article title into `n` collections, scored on retrieval only with Recall@5/@10, nDCG@5 and MAP. `python -m evaluations.datasets.mtrag_federated --config REF --n N --out PATH` builds the partition and emits the config that searches it.
### Removed
- `dot` removed from `search.vector_index_metric`; switch to `cosine` or `l2`
and rerun `create-index`.
### Fixed
- Batched evaluation ingest converts inline content as text instead of letting `HaikuRAG.convert` disambiguate it, so a passage beginning with a URL is stored rather than fetched over HTTP. 187 MTRAG cloud and fiqa passages start with one; no clapnq passage does, so no existing dataset's numbers change.
- `mtrag_federated` builds vacuum each collection after ingest and assert the chunks FTS index covers every row. Without the vacuum the index stays at zero rows, and full-text search returns near-arbitrary rows while still returning results.
### Added
- `evaluations run --retrieval-limit N`: candidates each database fetches during the retrieval benchmark, overriding the dataset's `retrieval_limit`.
- `mtrag_pooled` evaluation dataset and its reference config `evaluations/configs/mtrag_pooled.yaml`: all four MTRAG domains pooled and partitioned across `n` collections, `--alpha` interpolating between one domain per collection and a uniform shard.
- `mtrag_federated` evaluation dataset and its reference config `evaluations/configs/mtrag_federated.yaml`: MTRAG ClapNQ partitioned by article title into `n` collections, scored on retrieval only with Recall@5/@10, nDCG@5 and MAP. `python -m evaluations.datasets.mtrag_federated --config REF --n N --out PATH` builds the partition and emits the config that searches it.
### Fixed
- FTS and hybrid search on a database whose FTS index covers no rows. Chunk
writes now build the index and rebuild it if it covers none; `haiku-rag
vacuum` also repairs it.