A broad and consolidated ecosystem has developed around it for each application domain (Computer Vision, Natural Language Processing, Graph Neural Networks, Diffusion Models etc.).
Quantyca has confirmed in its projects a profound expertise in PyTorch and the surrounding frameworks, such as fast.ai, HuggingFace, Detectron and Timm.
- Timm and Detectron for Computer Vision (Classification, Detection, Segmentation etc.)
- HuggingFace for NLP (Text Analysis, Entity Extraction, Semantic Search)
- fast.ai as a meta-framework for prototyping, training and deployment
These libraries have two goals:
- Create high-level APIs to make it easy to design and deploy solutions
- Offer a catalogue of semi-processed models to enable fine-tuning on proprietary customer datasets