DeBERTa-v3-base-mnli-fever-anli
| Property | Value |
|---|---|
| Parameter Count | 184M |
| License | MIT |
| Paper | DeBERTa Paper |
| Training Data | 763,913 NLI pairs |
What is DeBERTa-v3-base-mnli-fever-anli?
This is a specialized version of Microsoft's DeBERTa-v3-base model, fine-tuned specifically for natural language inference (NLI) tasks. The model has been trained on a comprehensive dataset combining MultiNLI, Fever-NLI, and Adversarial-NLI, making it particularly robust for zero-shot classification tasks. Notably, it outperforms many larger models on the ANLI benchmark despite its base size.
Implementation Details
The model leverages the advanced DeBERTa-v3 architecture with significant improvements in pre-training objectives compared to previous versions. It was trained using mixed precision training with specific hyperparameters including a learning rate of 2e-05, batch size of 32, and 3 training epochs.
- Zero-shot classification capability with simple pipeline implementation
- Supports both single-label and multi-label classification
- Optimized for inference tasks with three output labels: entailment, neutral, and contradiction
Core Capabilities
- High accuracy on MNLI benchmark (90.3%)
- Strong performance on FEVER-NLI (77.7%)
- Competitive ANLI performance (49.5% on R3)
- Efficient zero-shot classification for various tasks
Frequently Asked Questions
Q: What makes this model unique?
The model combines DeBERTa-v3's advanced architecture with comprehensive NLI training, making it particularly effective for zero-shot classification while being more efficient than larger models.
Q: What are the recommended use cases?
The model excels at zero-shot text classification, natural language inference, and hypothesis-premise pair analysis. It's particularly useful when you need to classify text without task-specific training data.
