Investigate third parties tensor abstraction compatible with all machine learning frameworks
The goal of this issue is to investigate third parties libraries that provide a generic way to handle / manipulate tensors regardless of which machine learning one is using.
Such abstraction will then extend Vector
class that already provide an abstraction for handling vector / matrix/tensor computation regardless of the machine learning framework used
Promising frameworks to investigate:
- onnx
- ivy
- tensorly
Nota: this task requires completing Vector
class implementation #416
Edited by BOUILLARD Yannick