Skip to content

fix: fix search in memory with AnyEmbedding #1696

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
JoanFM merged 6 commits into from
Jul 12, 2023
Merged

fix: fix search in memory with AnyEmbedding #1696

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
a2ccb08
fix: fix inmemory search with anyembedding
Jul 10, 2023
bc0e40d
fix: allow find in in memory index with AnyEmbedding
Jul 10, 2023
e7315e8
test: add tests for torch and tf
Jul 11, 2023
2babe51
fix: try to pass all mypy checks
Jul 11, 2023
bab8099
test: undo changes in docker compose
Jul 11, 2023
5a05a17
Merge branch 'main' into anytensor-in-memory-search-fix
Jul 11, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions docarray/utils/find.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,8 @@
from docarray.helper import _get_field_type_by_access_path
from docarray.typing import AnyTensor
from docarray.typing.tensor.abstract_tensor import AbstractTensor
from docarray.typing.tensor.embedding import AnyEmbedding
from docarray.typing.tensor.embedding.ndarray import NdArrayEmbedding
from docarray.utils._internal._typing import safe_issubclass


Expand Down Expand Up @@ -207,6 +209,9 @@ class MyDocument(BaseDoc):
descending = metric.endswith('_sim') # similarity metrics are descending

embedding_type = _da_attr_type(index, search_field)
if embedding_type == AnyEmbedding:
embedding_type = NdArrayEmbedding

comp_backend = embedding_type.get_comp_backend()

# extract embeddings from query and index
Expand Down
14 changes: 14 additions & 0 deletions tests/index/in_memory/test_in_memory.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@
from torch import rand

from docarray import BaseDoc, DocList
from docarray.documents import TextDoc
from docarray.index.backends.in_memory import InMemoryExactNNIndex
from docarray.typing import NdArray, TorchTensor

Expand Down Expand Up @@ -112,6 +113,19 @@ class MyDoc(BaseDoc):
assert len(scores) == 0


def test_with_text_doc():
index = InMemoryExactNNIndex[TextDoc]()

docs = DocList[TextDoc](
[TextDoc(text='hey', embedding=np.random.rand(128)) for i in range(200)]
)
index.index(docs)
res = index.find_batched(docs[0:10], search_field='embedding')
assert len(res.documents) == 10
for r in res.documents:
assert len(r) == 5


def test_concatenated_queries(doc_index):
query = SchemaDoc(text='query', price=0, tensor=np.ones(10))

Expand Down