romfast/rar-autopass
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

perf+calitate sugestii k-NN: matvec numpy, vot top-5 cu prag calibrat, indicatori import

Browse Source 
- embeddings: corpus ca matrice numpy cu norme precalculate; suggest_nearest
  = un matvec (~0.6ms/query fata de ~500ms cosine pur-Python la 17k vectori)
- enrich_suggestions: vot ponderat cu similaritatea pe top-5 vecini (NUL =
  eticheta proprie); prag 0.5 -> 0.88, calibrat LOO pe corpusul SILVER
  (tools/mapare-llm/knn_calibrate.py): precizie 90.5% -> 93.1%, cod gresit
  preselectat 7.2% -> 4.7%; sub prag abtinere -> preselectie fuzzy
- UI: codul sugerat de sistem afisat explicit cu sursa si scorul, separat de
  lista fuzzy ("potrivire text"); indicator de progres reparat pe upload
  (display:inline anula .htmx-indicator) si adaugat pe pasii 2->3 si
  "Salveaza maparile"

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Claude Agent
2026-07-07 10:10:58 +00:00
parent 65b6e4694d
commit 809823a8ae
13 changed files with 358 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
app/web/routes.py
View File

@@ -1392,7 +1392,10 @@ def _nemapate_pentru_submission(row, nomenclator: list[dict], conn=None) -> list
"denumire": item.get("denumire"),
"suggestions": suggest_codes(item.get("denumire"), nomenclator, limit=5),
"sugestie_principala": None,
"surse_sugestie": {"gold_partajat": None, "silver": None, "embedding": None, "nul": False},
"surse_sugestie": {
"gold_partajat": None, "silver": None,
"embedding": None, "embedding_similaritate": None, "nul": False,
},
}
# L14-S6: imbogatire cu GOLD partajat > SILVER > embeddings (SUGGESTION-ONLY, #13)
if conn is not None:
@@ -3034,7 +3037,10 @@ def _collect_unmapped_ops(preview_rows: list[dict], nomenclator: list[dict], con
entry["suggestions"] = suggest_codes(entry["denumire"], nomenclator, limit=5)
# Init default pe FIECARE entry -> contract template identic (conn=None inclus).
entry["sugestie_principala"] = None
entry["surse_sugestie"] = {"gold_partajat": None, "silver": None, "embedding": None, "nul": False}
entry["surse_sugestie"] = {
"gold_partajat": None, "silver": None,
"embedding": None, "embedding_similaritate": None, "nul": False,
}
# L14-S6: imbogatire cu GOLD partajat > SILVER > embeddings (SUGGESTION-ONLY, #13)
if conn is not None:
enriched = enrich_suggestions(conn, entry["denumire"])