deberta-v3-large-absa-v1.1
| Property | Value |
|---|---|
| Author | yangheng |
| Base Model | microsoft/deberta-v3-large |
| Training Data | 180k+ ABSA examples |
| Paper | arXiv:2110.08604 |
What is deberta-v3-large-absa-v1.1?
This is a specialized DeBERTa-v3-large model fine-tuned for aspect-based sentiment analysis (ABSA). Built on the FAST-LCF-BERT architecture, it's trained on over 180,000 examples from various domains including restaurant reviews, laptop reviews, and social media content.
Implementation Details
The model is implemented using the transformer architecture and can be easily loaded using the Hugging Face transformers library. It's based on the FAST-LCF-BERT model architecture and uses microsoft/deberta-v3-large as its foundation.
- Trained on multiple ABSA datasets including SemEval, ACL Twitter, MAMS, and Yelp
- Optimized for common ABSA benchmark datasets like Laptop14 and Rest14
- Implements advanced sentiment analysis capabilities for specific aspects within text
Core Capabilities
- Fine-grained sentiment analysis at aspect level
- Handles multiple domains (restaurants, laptops, electronics, clothing)
- Supports both training and inference on standard ABSA tasks
- Advanced contextual understanding for aspect-sentiment relationships
Frequently Asked Questions
Q: What makes this model unique?
This model specializes in aspect-based sentiment analysis with extensive training on diverse datasets (30k+ samples). It's particularly notable for its ability to analyze sentiment for specific aspects of products or services rather than just overall sentiment.
Q: What are the recommended use cases?
The model is ideal for analyzing customer reviews, social media feedback, and any text where sentiment needs to be analyzed for specific aspects or features. It's particularly well-suited for applications in e-commerce, restaurant reviews, and product feedback analysis.
