Skip to content
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

fix faiss setup to take dataframe as input #378

Merged
merged 3 commits into from
Jul 1, 2023
Merged

fix faiss setup to take dataframe as input #378

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4248d21
fix faiss setup to take dataframe as input
jperez999 Jun 29, 2023
0762f75
use make df in faiss test
jperez999 Jun 30, 2023
e127ac8
Merge branch 'main' into fix-faiss-setup
jperez999 Jul 1, 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
16 changes: 12 additions & 4 deletions merlin/systems/dag/ops/faiss.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@
import numpy as np

from merlin.core.dispatch import HAS_GPU
from merlin.core.protocols import Transformable
from merlin.core.protocols import DataFrameLike, Transformable
from merlin.dag import ColumnSelector
from merlin.schema import ColumnSchema, Schema
from merlin.systems.dag.ops.operator import InferenceOperator
Expand Down Expand Up @@ -189,7 +189,13 @@ def validate_schemas(
)


def setup_faiss(item_vector, output_path: str, metric=faiss.METRIC_INNER_PRODUCT):
def setup_faiss(
item_vector: DataFrameLike,
output_path: str,
metric=faiss.METRIC_INNER_PRODUCT,
item_id_column="item_id",
embedding_column="embedding",
):
"""
Utiltiy function that will create a Faiss index from a set of embedding vectors

Expand All @@ -200,8 +206,10 @@ def setup_faiss(item_vector, output_path: str, metric=faiss.METRIC_INNER_PRODUCT
output_path : string
target output path
"""
ids = item_vector[:, 0].astype(np.int64)
item_vectors = np.ascontiguousarray(item_vector[:, 1:].astype(np.float32))
ids = item_vector[item_id_column].to_numpy().astype(np.int64)
item_vectors = np.ascontiguousarray(
np.stack(item_vector[embedding_column].to_numpy()).astype(np.float32)
)

index = faiss.index_factory(item_vectors.shape[1], "IVF32,Flat", metric)
index.nprobe = 8
Expand Down
9 changes: 6 additions & 3 deletions tests/unit/systems/ops/faiss/test_executor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@
import numpy as np
import pytest

from merlin.core.dispatch import make_df
from merlin.schema import ColumnSchema, Schema
from merlin.systems.dag.ensemble import Ensemble
from merlin.systems.dag.ops.faiss import QueryFaiss, setup_faiss
Expand Down Expand Up @@ -57,9 +58,11 @@ def test_faiss_in_triton_executor_model(tmpdir):
)

faiss_path = tmpdir / "faiss.index"
item_ids = np.arange(0, 100).reshape(-1, 1)
item_embeddings = np.ascontiguousarray(np.random.rand(100, 128))
setup_faiss(np.concatenate((item_ids, item_embeddings), axis=1), faiss_path)
item_ids = np.arange(0, 100)
item_embeddings = np.random.rand(100, 128)
# cannot turn a list column in cudf directly to numpy so must delegate to pandas as bridge
df = make_df({"item_id": item_ids, "embedding": item_embeddings.tolist()}, device="cpu")
setup_faiss(df, faiss_path)

request_schema = Schema(
[
Expand Down