Skip to content

Instantly share code, notes, and snippets.

@maropu

maropu/ComplExでベクトル化された関係の索引化（with Vertex AI Matching Engine）に関して

Created October 3, 2022 06:49
Show Gist options
  • Save maropu/b273f1883e96f4b3aa1f3fae91cc0822 to your computer and use it in GitHub Desktop.
Save maropu/b273f1883e96f4b3aa1f3fae91cc0822 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
ComplExでベクトル化された関係の索引化（with Vertex AI Matching Engine）に関して
- Complex Embeddings for Simple Link Prediction, https://arxiv.org/abs/1606.06357
- Vertex AI Matching Engine: https://cloud.google.com/vertex-ai/docs/matching-engine
@maropu
Copy link
Author

maropu commented Oct 3, 2022 •
edited
Loading

下記の[a_0, a_1, a_d-1, b_0, b_1, b_d-1]（右側のベクトル）をVertex AI Matching Engineで索引化、headのentityの入力（左側のベクトル）に対して近傍のベクトルを索引経由でtop-Kのentityを取得（注意: 索引するベクトルの次元が2倍になるため、それに応じて索引サイズが増大）

texclip20221003152519
texclip20221003152550
texclip20221003152725
texclip20221003152642

texclip20221003150825
texclip20221003150949

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment