phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-17 11:19:47 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

chore(migrate apis): move VectorDBWithIndex from embeddings to openai_embeddings

Browse source 
note: query_chunks has no unit test coverage
This commit is contained in:
Matthew Farrellee 2025-08-31 08:48:10 -04:00
parent b12cd528ef
commit 486cd8679a
2 changed files with 12 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

8
llama_stack/providers/utils/memory/vector_store.py
View file

@ -294,11 +294,11 @@ class VectorDBWithIndex:
_validate_embedding(c.embedding, i, self.vector_db.embedding_dimension)
if chunks_to_embed:
resp = await self.inference_api.embeddings(
resp = await self.inference_api.openai_embeddings(
self.vector_db.embedding_model,
[c.content for c in chunks_to_embed],
)
for c, embedding in zip(chunks_to_embed, resp.embeddings, strict=False):
for c, embedding in zip(chunks_to_embed, resp.data, strict=False):
c.embedding = embedding
embeddings = np.array([c.embedding for c in chunks], dtype=np.float32)
@ -334,8 +334,8 @@ class VectorDBWithIndex:
if mode == "keyword":
return await self.index.query_keyword(query_string, k, score_threshold)
embeddings_response = await self.inference_api.embeddings(self.vector_db.embedding_model, [query_string])
query_vector = np.array(embeddings_response.embeddings[0], dtype=np.float32)
embeddings_response = await self.inference_api.openai_embeddings(self.vector_db.embedding_model, [query_string])
query_vector = np.array(embeddings_response.data[0], dtype=np.float32)
if mode == "hybrid":
return await self.index.query_hybrid(
query_vector, query_string, k, score_threshold, reranker_type, reranker_params

14
tests/unit/rag/test_vector_store.py
View file

@ -218,11 +218,13 @@ class TestVectorDBWithIndex:
Chunk(content="Test 2", embedding=None, metadata={}),
]
mock_inference_api.embeddings.return_value.embeddings = [[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]]
mock_inference_api.openai_embeddings.return_value.data = [[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]]
await vector_db_with_index.insert_chunks(chunks)
mock_inference_api.embeddings.assert_called_once_with("test-model without embeddings", ["Test 1", "Test 2"])
mock_inference_api.openai_embeddings.assert_called_once_with(
"test-model without embeddings", ["Test 1", "Test 2"]
)
mock_index.add_chunks.assert_called_once()
args = mock_index.add_chunks.call_args[0]
assert args[0] == chunks
@ -246,7 +248,7 @@ class TestVectorDBWithIndex:
await vector_db_with_index.insert_chunks(chunks)
mock_inference_api.embeddings.assert_not_called()
mock_inference_api.openai_embeddings.assert_not_called()
mock_index.add_chunks.assert_called_once()
args = mock_index.add_chunks.call_args[0]
assert args[0] == chunks
@ -288,7 +290,7 @@ class TestVectorDBWithIndex:
with pytest.raises(ValueError, match="has dimension 4, expected 3"):
await vector_db_with_index.insert_chunks(chunks_wrong_dim)
mock_inference_api.embeddings.assert_not_called()
mock_inference_api.openai_embeddings.assert_not_called()
mock_index.add_chunks.assert_not_called()
async def test_insert_chunks_with_partially_precomputed_embeddings(self):
@ -308,11 +310,11 @@ class TestVectorDBWithIndex:
Chunk(content="Test 3", embedding=None, metadata={}),
]
mock_inference_api.embeddings.return_value.embeddings = [[0.1, 0.1, 0.1], [0.3, 0.3, 0.3]]
mock_inference_api.openai_embeddings.return_value.data = [[0.1, 0.1, 0.1], [0.3, 0.3, 0.3]]
await vector_db_with_index.insert_chunks(chunks)
mock_inference_api.embeddings.assert_called_once_with(
mock_inference_api.openai_embeddings.assert_called_once_with(
"test-model with partial embeddings", ["Test 1", "Test 3"]
)
mock_index.add_chunks.assert_called_once()