multilingual-MiniLMv2-L6-mnli-xnli
| Property | Value |
|---|---|
| Parameter Count | 107M |
| License | MIT |
| Languages Supported | 100+ (15 primary languages) |
| Primary Paper | Link |
| Average XNLI Accuracy | 71.3% |
What is multilingual-MiniLMv2-L6-mnli-xnli?
This is a powerful multilingual model designed for natural language inference (NLI) and zero-shot classification. Developed by Microsoft and fine-tuned by MoritzLaurer, it represents a distilled version of XLM-RoBERTa-large, optimized for efficiency while maintaining strong performance across multiple languages.
Implementation Details
The model was trained on the XNLI development dataset and MNLI train dataset, encompassing 392,702 texts in total. It employs a 6-layer architecture with specific training parameters including a learning rate of 4e-05 and batch size of 64.
- Supports zero-shot classification across 100+ languages
- Achieves 71.3% average accuracy on XNLI test set
- Processes 6000+ texts per second on A100 GPU
- Optimized for both speed and memory efficiency
Core Capabilities
- Multilingual Natural Language Inference
- Zero-shot Classification
- Cross-lingual Transfer Learning
- High-speed Inference Processing
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its excellent balance between performance and efficiency, being significantly smaller than its teacher model while maintaining strong multilingual capabilities. Its 6-layer architecture makes it particularly suitable for production environments where speed is crucial.
Q: What are the recommended use cases?
The model is ideal for zero-shot classification tasks across multiple languages, natural language inference, and scenarios requiring quick inference times. It's particularly effective for applications needing multilingual support without the computational overhead of larger models.
