Text Classification Models
Knowledgator provides a collection of powerful models for text classification tasks, supporting both zero-shot and few-shot learning scenarios. These models are designed to handle a variety of use cases such as sentiment analysis, topic detection, intent classification, and more.
Available Models​
🔹 ComprehendIt​
A DeBERTaV3-based model fine-tuned on NLI and classification datasets. Suitable for general-purpose zero-shot classification across multiple domains.
🔹 GLiClass​
A lightweight model optimized for high-speed inference in low-resource environments, suitable for real-time applications and edge devices.
Key Features​
- Support for multi-label and multi-class classification.
- Optimized for zero-shot inference using NLI-style input formulation.
- Fine-tuning capabilities with the LiqFit framework.
- Usable with Hugging Face's
pipelineAPI or custom training setups.
Use Cases​
- Sentiment Analysis
- Intent Recognition
- Topic Classification
- Content Moderation
- Customer Feedback Analysis
For detailed documentation and usage examples, visit Knowledgator Docs.