About |
With its groundbreaking research in ML, we aim to overcome the limitations of large-scale generative models by focusing on resource efficiency and task-specific precision. Our commitment to promoting effective approaches with encoder-based models to deliver versatile ML solutions for diverse information extraction tasks, which can adapt to different tasks quickly with minimal data and effort.
We have released two ML models: 'Comprehend-it', an open-source text classification model, and 'Salamandra', a prompt-based token-classification model. These models are designed to address a broad spectrum of popular information extraction tasks.
Models are available for download on Hugging Face, users can fine-tune and deploy these models seamlessly on their systems or access them via API.