Yiorgis Gozadinos
46b8fa3fef
vb
2026-06-08 12:24:04 +03:00
Yiorgis Gozadinos
7941c6738c
Merge pull request #427 from ggozad/feat/t2-ragbench
...
Add T²-RAGBench evaluation
2026-06-08 12:15:06 +03:00
Yiorgis Gozadinos
9ef25e2e53
Order benchmarks docs ORB, T², Wix
2026-06-08 11:51:21 +03:00
Yiorgis Gozadinos
295570797c
Disable Logfire scrubbing in evaluations
2026-06-08 11:51:13 +03:00
Yiorgis Gozadinos
e9bd56467c
Add T²-RAGBench leaderboard submission exporter
2026-06-08 11:51:02 +03:00
Yiorgis Gozadinos
a3a73f1331
Add --filter-ids to run QA on a case-id subset
2026-06-08 09:36:01 +03:00
Yiorgis Gozadinos
d386d7f900
Bump haiku.skills to 0.17.2
2026-06-06 14:52:05 +03:00
Yiorgis Gozadinos
cd77bd9889
Bound analysis execute_code calls to avoid request-limit nulls
2026-06-06 14:52:05 +03:00
Yiorgis Gozadinos
1d7d540270
Match the numeric scale convention in Number-Match
2026-06-06 14:52:05 +03:00
Yiorgis Gozadinos
d5b0aafa2b
Add T²-RAGBench TAT-DQA subset; generalize subset layout
2026-06-06 14:52:05 +03:00
Yiorgis Gozadinos
de9731d5f3
Score Number-Match on the declared ANSWER line, magnitude and ×100 scale
2026-06-06 14:52:05 +03:00
Yiorgis Gozadinos
390deb4203
Normalize unicode signs
2026-06-06 14:52:04 +03:00
Yiorgis Gozadinos
80594cd38c
Handle percent/decimal convention in Number-Match
2026-06-06 14:52:04 +03:00
Yiorgis Gozadinos
c2a48b55c0
Add deterministic Number-Match QA scoring for T²-RAGBench
2026-06-06 14:52:04 +03:00
Yiorgis Gozadinos
9f1bd9940a
Add T²-RAGBench FinQA evaluation dataset
2026-06-06 14:52:04 +03:00
Yiorgis Gozadinos
3807c48a60
Merge pull request #425 from ggozad/fix/sandbox-borrowed-bug
...
Fix analysis-sandbox "Already borrowed" crash under concurrent tool calls
2026-06-06 14:16:29 +03:00
Yiorgis Gozadinos
52738cbc0b
Always create the shared-connection lock so serialization can't be skipped
2026-06-06 10:51:07 +03:00
Yiorgis Gozadinos
4a65a69a27
Make the shared-connection lock optional for direct tool calls
2026-06-05 18:04:56 +03:00
Yiorgis Gozadinos
e59ee56002
Serialize shared-connection access across skill tools and the sandbox
2026-06-05 17:22:11 +03:00
Yiorgis Gozadinos
b47583258e
Run analysis sandbox VFS reads on the calling loop via the skill connection
2026-06-05 16:20:11 +03:00
Yiorgis Gozadinos
e0a892ec97
Update embedding-drift docs for read-only opens
2026-06-05 12:14:14 +03:00
Yiorgis Gozadinos
916b84bff4
vb
2026-06-05 12:09:57 +03:00
Yiorgis Gozadinos
f3cc1e8c7c
Merge pull request #422 from ggozad/fix/sandbox-lancedb-event-loop
...
Fix analysis sandbox "Already borrowed" crash on document VFS reads
2026-06-05 12:08:01 +03:00
Yiorgis Gozadinos
7906688bf1
Fix analysis sandbox "Already borrowed" crash on VFS reads
2026-06-05 11:50:19 +03:00
Yiorgis Gozadinos
a2def043cc
Merge pull request #421 from ggozad/fix/read-only-update
...
Read operations no longer modify the database
2026-06-05 11:46:46 +03:00
Yiorgis Gozadinos
abefdcdd41
Test vector_dim mismatch raises in read-only mode
2026-06-05 10:55:38 +03:00
Yiorgis Gozadinos
414be551fb
Open read CLI verbs read-only
2026-06-05 10:48:49 +03:00
Yiorgis Gozadinos
0e3d53791f
Add rebuild --set-embedder to reconcile embedder identity
2026-06-05 10:42:15 +03:00
Yiorgis Gozadinos
213569601b
Stop rewriting stored embedding settings on database open
2026-06-05 10:35:50 +03:00
Yiorgis Gozadinos
b6bbea3d64
Stop writing the schema version on database open
2026-06-05 10:32:30 +03:00
Yiorgis Gozadinos
e70f271ac7
vb
2026-06-04 14:13:55 +03:00
Yiorgis Gozadinos
c688871b61
Merge pull request #418 from ggozad/feat/ingester-postgres-queue
...
Support a database server for the ingester queue (SQLite + Postgres)
2026-06-04 14:07:14 +03:00
Yiorgis Gozadinos
e2e0a8dc1b
Cover the Postgres queue construction paths in CI
2026-06-04 12:21:47 +03:00
Yiorgis Gozadinos
73e8ac2dd6
Build the SQLite queue URL without reparsing the path
2026-06-04 12:14:27 +03:00
Yiorgis Gozadinos
a04dc9445b
Pre-ping the Postgres queue engine
2026-06-04 11:29:31 +03:00
Yiorgis Gozadinos
e37d764ab2
Add a docker-compose for integration test services
2026-06-04 11:13:16 +03:00
Yiorgis Gozadinos
1717bd4996
Make the SQLite job claim atomic across processes
2026-06-03 16:57:57 +03:00
Yiorgis Gozadinos
5cc32f111e
Mask the dburi password in queue init/migrate output
2026-06-03 14:42:32 +03:00
Yiorgis Gozadinos
a3cc13230f
test the Postgres queue and document dburi
2026-06-03 14:34:09 +03:00
Yiorgis Gozadinos
44089e5b1f
Support a dburi for the ingester queue (SQLite + Postgres)
...
Migrate the ingester queue storage from raw aiosqlite to SQLAlchemy Core
async. The backend is chosen by ingester.queue.dburi: a SQLAlchemy async
URL points the queue at a database server, and SQLite remains the default
when unset. The Postgres path claims jobs with FOR UPDATE SKIP LOCKED so
multiple ingester processes can share one queue; SQLite caps the pool to a
single connection to keep the select-then-update claim atomic.
2026-06-03 14:34:09 +03:00
Yiorgis Gozadinos
4eae86225e
vb
2026-06-03 14:26:58 +03:00
Yiorgis Gozadinos
0b58e6dc14
Merge pull request #417 from ggozad/fix/atomic-rename-race
...
skip spurious ingester DELETE when the resource is back on its source
2026-06-03 14:17:51 +03:00
Yiorgis Gozadinos
8afdd46176
skip spurious ingester DELETE when the resource is back on its source
2026-06-03 14:05:53 +03:00
Yiorgis Gozadinos
5b879aa6ed
Merge pull request #416 from ggozad/fix/rebuild-embed-images
...
Fix rebuild --embed-only corrupting picture embeddings
2026-06-03 13:49:04 +03:00
Yiorgis Gozadinos
a65757807e
Use cached HuggingFace models offline in test job
...
The Qwen tokenizer and cross-encoder pre-downloads call the HF metadata
API to revalidate even on a cache hit; a 429 there propagates instead of
falling back to the cached files, failing CI on HF throttling.
Skip the pre-download steps when the cache is restored and run pytest
with HF_HUB_OFFLINE/TRANSFORMERS_OFFLINE on a hit, so cached models are
used without any network revalidation; allow online on a miss so a fresh
cache key still populates. Rename the cache key so the snapshot
re-populates with every test model (the old key predated the
cross-encoder step and never cached it).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 12:11:33 +03:00
Yiorgis Gozadinos
651b22ddcf
Fix rebuild --embed-only corrupting picture embeddings
2026-06-03 11:46:43 +03:00
Yiorgis Gozadinos
6d95fbe74a
Merge pull request #414 from ggozad/feat/prune-ingestor-pool
...
Add retention window to ingester queue
2026-06-03 11:06:44 +03:00
Yiorgis Gozadinos
a423a6dd9c
authenticate HuggingFace downloads in test job
2026-06-03 10:57:45 +03:00
Yiorgis Gozadinos
46747d369a
Add retention window to ingester queue, prune terminal job rows past retention window
2026-06-03 10:38:58 +03:00
Yiorgis Gozadinos
83b7e37855
Merge pull request #413 from ggozad/feat/analysis-citation-improvement
...
Improve citation rate for analysis skill in lesser models (gemma4)
2026-06-03 10:03:54 +03:00
