PyTorch 2.0 is out! This major release upgrade brings about many new features, but the main improvements are under the hood.
1/6
1/6
The three main principles behind PyTorch
1. High-Performance eager execution
2. Pythonic internals
3. Good abstractions for Distributed, Autodiff, Data loading, Accelerators, etc.
PyTorch 2.0 is fully backward compatible with the previous versions of PyTorch.
2/6
1. High-Performance eager execution
2. Pythonic internals
3. Good abstractions for Distributed, Autodiff, Data loading, Accelerators, etc.
PyTorch 2.0 is fully backward compatible with the previous versions of PyTorch.
2/6
The main new feature is torch.compile, "a feature that pushes PyTorch performance to new heights and starts the move for parts of PyTorch from C++ back into Python."
3/6
3/6
This new feature can bring about 1.5 - 2.0X improvements in training times when tested on the newer server class NVIDIA GPUs.
4/6
4/6
The efficiency tests have been done on the following tasks:
* 46 models from HuggingFace Transformers
* 61 models from TIMM: a collection of state-of-the-art PyTorch image models by Ross Wightman
* 56 models from TorchBench: a curated set of popular code-bases
5/6
* 46 models from HuggingFace Transformers
* 61 models from TIMM: a collection of state-of-the-art PyTorch image models by Ross Wightman
* 56 models from TorchBench: a curated set of popular code-bases
5/6
Read the full release notes here: pytorch.org/get-started/py…
#MachineLearning #DeepLearning #ArtificialIntelligence #ML #DL #AI
6/6
#MachineLearning #DeepLearning #ArtificialIntelligence #ML #DL #AI
6/6
• • •
Missing some Tweet in this thread? You can try to
force a refresh
