pyTorch and onnxruntime backends and more
Created by: n-berezina-nn
- support of pyTorch large models: kandinsky, falcon-7B, gpt-2, etc
- support of PyTorch compile backends
- support of onnxruntime tensor-RT backend
- support of validation for TF-lite c++ launcher
- support of random inputs instead of datasets
- separate binary for tf-lite GPU delegate