jina-ai/mlx-retrieval

Train embedding and reranker models for retrieval tasks on Apple Silicon with MLX

37
/ 100
Emerging

Implements LoRA fine-tuning with contrastive losses (InfoNCE, NT-Xent) and hard negative mining, leveraging MLX Data for efficient streaming from local JSONL or Elasticsearch sources. Integrates with MTEB for evaluation and Weights & Biases for experiment tracking, supporting gradient accumulation to simulate large batch sizes on resource-constrained Apple Silicon hardware. Uses query/document prompt tokens with mean pooling for embedding generation, mirroring Jina's v3/v4 architectures.

177 stars. No commits in the last 6 months.

Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 10 / 25
Maturity 15 / 25
Community 10 / 25

How are scores calculated?

Stars

177

Forks

10

Language

Python

License

Apache-2.0

Last pushed

Sep 18, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/embeddings/jina-ai/mlx-retrieval"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.