mlx-embeddings-lora 1.0.2

Train Embedding Models on Apple silicon with MLX and the Hugging Face Hub

MLX-Embeddings-LoRA

With MLX-Embeddings-LoRA you can, train embedding models locally on Apple Silicon using MLX. Built on top of mlx-embeddings, supporting all models available in that package with contrastive learning algorithms optimized for semantic search, retrieval, and similarity tasks. Including:

• Qwen3
• XLM-RoBERTa
• BERT
• ModernBERT

Features

• 🚀 Efficient Training Methods

  • LoRA: Low-Rank Adaptation for efficient fine-tuning
  • DoRA: Weight-Decomposed Low-Rank Adaptation
  • Full-precision: Train all model parameters
  • Quantized training: QLoRA with 4-bit, 6-bit, or 8-bit quantization

• 📊 Contrastive Learning Algorithms

  • InfoNCE Loss: Temperature-scaled contrastive loss with in-batch negatives
  • Multiple Negatives Ranking Loss: Efficient ranking with batch negatives
  • Triplet Loss: Margin-based triplet optimization
  • NT-Xent Loss: Normalized temperature-scaled cross entropy (SimCLR-style)

So far only Text based embedding models and contrastive learning is supported, more features and algorythms are to come.

• 🔧 Flexible Dataset Support

  • Hugging Face datasets
  • JSONL files
  • Optional negative examples (auto-generated from batch if not provided)

• ⚡ Apple Silicon Optimized

  • Native MLX acceleration
  • Memory-efficient training
  • Gradient accumulation support

Installation

pip install -U mlx-embeddings-lora

Quick Start

Basic Training

mlx_embeddings_lora.train \
  --model mlx-community/all-MiniLM-L6-v2-4bit \
  --train \
  --data mlx-community/sentence-compression \
  --iters 600

With Configuration File

mlx_embeddings_lora.train --config config.yaml

Command-line flags will override corresponding values in the config file.

Dataset Format

Your dataset should contain anchor-positive pairs:

JSONL Format

{"anchor": "How do I reset my password?", "positive": "What's the process for password recovery?", "negative": "What's the weather today?"}
{"anchor": "Python tutorial for beginners", "positive": "Learn Python basics step by step"}
{"anchor": "Machine learning introduction", "positive": "Getting started with ML", "negative": "JavaScript frameworks overview"}

Note: The negative field is optional. If not provided, the training algorithm will automatically use in-batch negatives from other examples in the batch.

Key Parameters

Training Method

• --train-type: Choose training method
  • lora (default): Low-Rank Adaptation
  • dora: Weight-Decomposed Low-Rank Adaptation
  • full: Full parameter fine-tuning

LoRA Configuration

• --lora-rank: Rank of LoRA matrices (default: 16)
• --lora-alpha: LoRA scaling factor (default: 32)
• --lora-dropout: Dropout probability (default: 0.05)

Quantization

• --quantize: Enable quantized training (QLoRA)
• --quantize-bits: Quantization bits (4, 6, or 8)

Loss Function

• --loss-type: Contrastive loss algorithm
  • infonce: InfoNCE with temperature scaling (recommended)
  • mnr: Multiple Negatives Ranking Loss
  • triplet: Triplet loss with margin
  • nt_xent: NT-Xent (SimCLR-style)

Training Hyperparameters

• --batch-size: Training batch size (default: 32)
• --learning-rate: Learning rate (default: 5e-5)
• --iters: Number of training iterations (default: 1000)
• --max-seq-length: Maximum sequence length (default: 512)
• --gradient-accumulation-steps: Accumulate gradients over multiple steps

Core Training Parameters

# Model and data
--model <model_path>              # Model path or HF repo
--data <data_path>                # Dataset path or HF dataset name
--train-type lora                 # lora, dora, or full
--train-mode infonce              # infonce, mnr, triplet, nt_xent

# Training schedule
--batch-size 4                    # Batch size
--iters 1000                      # Training iterations
--epochs 3                        # Training epochs (ignored if iters set)
--learning-rate 1e-5              # Learning rate
--gradient-accumulation-steps 1   # Gradient accumulation

# Model architecture
--num-layers 16                   # Layers to fine-tune (-1 for all)
--max-seq-length 2048            # Maximum sequence length

# LoRA parameters
--lora-parameters '{"rank": 8, "dropout": 0.0, "scale": 10.0}'

# Optimization
--optimizer adam                  # adam, adamw, qhadam, muon
--lr-schedule cosine             # Learning rate schedule
--grad-checkpoint                # Enable gradient checkpointing

# Quantization
--load-in-4bits                  # 4-bit quantization
--load-in-6bits                  # 6-bit quantization  
--load-in-8bits                  # 8-bit quantization

# Monitoring
--steps-per-report 10            # Steps between loss reports
--steps-per-eval 200             # Steps between validation
--val-batches 25                 # Validation batches (-1 for all)
--wandb project_name             # WandB logging

# Checkpointing
--adapter-path ./adapters        # Save/load path for adapters
--save-every 100                 # Save frequency
--resume-adapter-file <path>     # Resume from checkpoint
--fuse                           # Fuse and save trained model

Advanced Features

Automatic Negative Sampling

If your dataset doesn't include negative examples, the training will automatically use in-batch negatives:

{"anchor": "Query 1", "positive": "Relevant doc 1"}
{"anchor": "Query 2", "positive": "Relevant doc 2"}
{"anchor": "Query 3", "positive": "Relevant doc 3"}

For each anchor, positives from other examples in the batch serve as negatives.

Gradient Accumulation

For larger effective batch sizes with limited memory:

mlx_embeddings_lora.train \
  --model your-model \
  --batch-size 16 \
  --gradient-accumulation-steps 4  # Effective batch size: 64

Model Export

After training, export your fine-tuned model and upload to Hugging Face:

mlx_embeddings_lora.export \
  --model ./output/checkpoint-1000 \
  --output ./my-finetuned-model \
  --repo username/model-name

Performance Tips

1. Start with LoRA: More memory efficient than full fine-tuning
2. Use in-batch negatives: Skip explicit negatives for efficiency
3. Tune temperature: Lower (0.05-0.07) for harder negatives, higher (0.1-0.2) for softer
4. Batch size: Larger batches = more negatives = better performance
5. Gradient accumulation: Increase effective batch size without OOM
6. QLoRA for large models: Use 4-bit quantization for models >1B parameters

Citation

If you use mlx-embeddings-lora in your research, please cite:

@software{mlx_embeddings_lora,
  title = {mlx-embeddings-lora: Efficient Embedding Model Training on Apple Silicon},
  author = {Gökdneiz Gülmez},
  year = {2025},
  url = {https://github.com/Goekdeniz-Guelmez/mlx-embeddings-lora}
}

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE for details.

Acknowledgments

• Built on MLX by Apple
• Extends mlx-embeddings
• Inspired by Sentence-Transformers