April 19, 2023
Accelerating Large Language Models with Accelerated Transformers
TL;DR. We show how to use Accelerated PyTorch 2.0 Transformers and the newly introduced torch.compile() method to accelerate Large Language Models on the example of nanoGPT, a compact open-source implementation of the GPT model from Andrej Karpathy. Using the new scaled dot product attention operator introduced with Accelerated PT2 Transformers, we select the flash_attention custom kernel and achieve faster training time per batch (measured with Nvidia A100 GPUs), going from a ~143ms/batch ba...
April 15, 2023
Experience the power of PyTorch 2.0 on AMD Solutions
PyTorch 2.0 represents a significant step forward for the PyTorch machine learning framework. The stable release of PyTorch 2.0 brings new features that unlock even higher performance, while remaining backward compatible with prior releases and retaining the Pythonic focus which has helped to make PyTorch so enthusiastically adopted by the AI/ML community. AMD has long been a strong proponent of PyTorch, and we are delighted that the PyTorch 2.0 stable release includes support for AMD Instin...
April 14, 2023
Accelerated Generative Diffusion Models with PyTorch 2
TL;DR: PyTorch 2.0 nightly offers out-of-the-box performance improvement for Generative Diffusion models by using the new torch.compile() compiler and optimized implementations of Multihead Attention integrated with PyTorch 2.
April 07, 2023
Straggler Mitigation On PyTorch DDP By Hierarchical SGD
PyTorch DDP has been widely adopted across the industry for distributed training, which by default runs synchronous SGD to synchronize gradients across model replicas at every step. The performance of this technique is critical for fast iteration during model exploration as well as resource and cost saving. The performance is critical for fast iteration and cost saving of model development and exploration. To resolve a ubiquitous performance bottleneck introduced by slow nodes in large-scale ...
April 07, 2023
Celebrate PyTorch 2.0 with New Performance Features for AI Developers
Congratulations to the PyTorch Foundation for its release of PyTorch 2.0! In this blog, I discuss the four features for which Intel made significant contributions to PyTorch 2.0:
April 03, 2023
PyTorch & OpenXLA: The Path Forward
As we celebrate the release of OpenXLA, PyTorch 2.0, and PyTorch/XLA 2.0, it’s worth taking a step back and sharing where we see it all going in the short to medium term. With PyTorch adoption leading in the AI space and XLA supporting best-in-class compiler features, PyTorch/XLA is well positioned to provide a cutting edge development stack for both model training and inference. To achieve this, we see investments in three main areas:
March 28, 2023
Accelerated PyTorch 2 Transformers
The PyTorch 2.0 release includes a new high-performance implementation of the PyTorch Transformer API with the goal of making training and deployment of state-of-the-art Transformer models affordable. Following the successful release of “fastpath” inference execution (“Better Transformer”), this release introduces high-performance support for training and inference using a custom kernel architecture for scaled dot product attention (SPDA).
