Hi,
In the README.md for Real-time Detection Challenges, we can find the following information:
"_As such, users can use arbitrary versions of TensorFlow or PyTorch for their submissions with ease. In particular, we have tested TensorFlow 2.3.0 and 2.4.1, PyTorch 1.3 and 1.7.1, CUDA 10.0 and 11.0, and cuDNN 7 and 8, but other versions, especially older versions, should likely work with little issue. The only required dependency that users must have installed is numpy."
So, it is allowed to use different ML framework than TensorFlow and PyTorch?
Hi, In the README.md for Real-time Detection Challenges, we can find the following information:
"_As such, users can use arbitrary versions of TensorFlow or PyTorch for their submissions with ease. In particular, we have tested TensorFlow 2.3.0 and 2.4.1, PyTorch 1.3 and 1.7.1, CUDA 10.0 and 11.0, and cuDNN 7 and 8, but other versions, especially older versions, should likely work with little issue. The only required dependency that users must have installed is numpy."
So, it is allowed to use different ML framework than TensorFlow and PyTorch?