BGE-M3-Banking77-LoRA

This is a fine-tuned version of the BAAI/bge-m3 model specifically optimized for banking domain intent classification using the Banking77 dataset.

Model Description

• Base Model: BAAI/bge-m3
• Fine-tuning Method: LoRA (Low-Rank Adaptation)
• Domain: Banking and Financial Services
• Task: Intent Classification / Sentence Embedding
• Dataset: Banking77 (77 banking intents)

Performance

The model was evaluated on the Banking77 test set with the following results:

Metric	Score
Coverage@1	0.6666
Recall@1	0.6666
Precision@1	0.6898
F1 Score@1	0.6534
Accuracy@1	0.6666

Training Details

• Training Data: 10,003 examples from Banking77
• Test Data: 3,080 examples across 77 banking intents
• LoRA Rank: 16
• LoRA Alpha: 32
• Learning Rate: 2e-5
• Batch Size: 16 (with gradient accumulation)
• Epochs: 15
• Mixed Precision: FP16

Citation

If you use this model in your research, please cite:

@misc{bge-m3-banking77-lora,
  title={BGE-M3 Banking77 LoRA: Fine-tuned Banking Intent Classification},
  author={Sai Vamshi Atukuri},
  year={2025},
  url={https://huggingface.co/YOUR_USERNAME/bge-m3-banking77-lora}
}

License

This model is licensed under the MIT License, same as the base BGE-M3 model.