q deb422d03e Port v5-revert params from Lotus (best score 0.5517 vs our 0.5094) ... Score formula: recall×0.8 + ndcg×0.2 → recall 4x more important index: CHUNK_SIZE 256→384 (sweet spot, not too small, not too large) search: DENSE 80→50, SPARSE 200→150, RETRIEVE 150→100, RERANK 10→15 Fewer candidates = less noise = better recall Lotus experiments confirmed: 80/200/150 limits HURT vs 50/150/100. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>		2026-04-18 20:28:35 +03:00
..
__init__.py	Refactor index and search services	2026-04-18 13:57:15 +03:00
chunking.py	Filter is_system and is_hidden messages in chunking (align with Lotus reference)	2026-04-18 17:02:11 +03:00
cleaning.py	Refactor index and search services	2026-04-18 13:57:15 +03:00
Dockerfile	Port index and search logic from working Lotus reference implementation	2026-04-18 18:32:41 +03:00
index_schemas.py	Add logviewer project and fix Docker imports	2026-04-18 15:26:11 +03:00
main.py	Port v5-revert params from Lotus (best score 0.5517 vs our 0.5094)	2026-04-18 20:28:35 +03:00
Makefile	Fix date_range filter: use DatetimeRange for RFC3339 strings; set TEAM_ID=35230 in Makefiles, add release target	2026-04-18 16:26:14 +03:00
rendering.py	Add logviewer project and fix Docker imports	2026-04-18 15:26:11 +03:00
requirements.txt	Source VK code	2026-04-18 10:51:43 +03:00
sparse.py	Add logviewer project and fix Docker imports	2026-04-18 15:26:11 +03:00