Skip to content

GPU based embedding support #25

@Tonemon

Description

@Tonemon

Make the embedding service detect a CUDA device and load the model on GPU when available (device="cuda" in the FastEmbed/SentenceTransformer constructor). No code change needed on the Qdrant side. Reduces ingestion time significantly for large corpora.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions