feat(embeddings): every vector records the model whose space it lives in (#4132)
CI & Build / Python lint (push) Successful in 3s
CI & Build / Plugin hooks (push) Successful in 14s
CI & Build / TypeScript typecheck (push) Successful in 54s
CI & Build / integration (push) Successful in 1m1s
CI & Build / Python tests (push) Successful in 1m40s
CI & Build / Build & push image (push) Canceled after 27s

The four embedding tables stamped chunker_version but not the model, and
vector(384) is a width, not an identity: a same-width model swap would
write a second geometry beside the first with no error.

- embedding_model on note/rule/milestone/system embeddings (0109; existing
  rows stamped with the only model any install has ever run).
- Every write stamps EMBEDDING_MODEL; every backfill's "current" test is
  is_current_stamp(), both halves of calibration_stamp().
- migrate_floor refuses while any row its surface searches is off the live
  model, before sampling: re-embed, then migrate.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
2026-09-23 19:02:30 -04:00
co-authored by Claude Opus 5.5
parent 11b286d786
commit f4e9cd429b
10 changed files with 178 additions and 10 deletions
+15
View File
@@ -228,6 +228,7 @@ async def test_upsert_stores_one_versioned_row_per_chunk():
assert [r.chunk_index for r in rows] == list(range(len(chunks)))
assert [r.chunk_text for r in rows] == chunks
assert {r.chunker_version for r in rows} == {emb.CHUNKER_VERSION}
assert {r.embedding_model for r in rows} == {emb.EMBEDDING_MODEL}
assert {r.user_id for r in rows} == {42}
session.execute.assert_awaited() # the delete that makes replacement atomic
@@ -314,3 +315,17 @@ async def test_backfill_reembeds_notes_with_a_stale_chunker_version():
embedded = [call.args[0] for call in upsert.call_args_list]
assert embedded == [2], "only the stale note is re-embedded"
def test_a_row_is_current_only_in_the_live_models_space():
"""#4132: the column is a width, not an identity, so a same-width model
swap writes a second geometry with no error. The backfill's "current"
predicate has to name BOTH halves of the stamp, on every embedding table."""
from scribe.models.embedding import (
MilestoneEmbedding, NoteEmbedding, RuleEmbedding, SystemEmbedding,
)
from scribe.services import embeddings as emb
for table in (NoteEmbedding, RuleEmbedding, MilestoneEmbedding, SystemEmbedding):
clause = str(emb.is_current_stamp(table))
assert "chunker_version" in clause and "embedding_model" in clause, table