PyTorch Conference 2024

Arm is excited to be at the center of the world's largest compute ecosystem at the dawn of the AI era. A key tenant of our mission is to democratize AI capabilities, empowering millions of developers to put advanced AI features into the hands of billions of users.

In this presentation, we'll explore how Arm is enabling the world’s leading open-source AI frameworks to leverage power-efficient Arm-based computing platforms and Arm architecture features, as a tool for enabling fast and secure AI workloads. The session focuses on how our strategic partnership with the Pytorch and Executorch community is enabling a seamless and transparent developer experience, to run workloads everywhere from cloud to edge. This session will highlight some of our optimized libraries, upstreamed contributions and a wealth of AI-related developer material to build the future of AI on Arm.

Explore how the `tranformers` library grows and adapts to the fast paced and ever-changing AI field to bring the best to the AI community

This talk explores the journey of enabling LLMs in the PyTorch ecosystem, as well as how the teams behind AOT Inductor, ExecuTorch, and torchao collaborated to create torchchat, a showcase of PyTorch’s ability to run LLM inference everywhere.

Torchchat demonstrates the ubiquity, simplicity, and quality of PyTorch’s LLM support through performant, reproducible implementations for not only Python environments, but on desktop, server, and on-device as-well.

All of our work is open source and available on GitHub.

Mixture-of-Experts MoE (models) are becoming an increasingly popular architecture choice for large language models (LLMs). In this talk, we describe how to train MoE models with PyTorch. After discussing various performance tradeoffs, we use PyTorch distributed tools like DTensor to build custom parallelism approaches, including expert parallelism via MegaBlocks. We then show how to get near linear scaling to thousands of GPUs, combining PyTorch FSDP and HSDP with our parallelism strategies. We discuss many of the challenges of training at scale, including communication bottlenecks, hardware failures, and networking challenges. We further improve training at scale setups using tools like PyTorch Distributed Checkpointing for rapid saving and loading. We then highlight further optimizations to minimize challenges only present at scale, such as object store failures for large checkpoints.

At roboMUA we're leading the charge in building predictive AI models for diverse skin shades with the use of Convolutional Neural Networks (CNNs), and harnessing the power of Generative Adversarial Networks (GANs) specifically for generating realistic images of black hairstyles. Our session showcases PyTorch's versatility in both predictive and generative tasks, offering a comprehensive approach to inclusive AI. For predictive AI models, we leverage PyTorch's flexible framework to develop CNNs. Through innovative techniques in feature engineering and model architecture design, we demonstrate how PyTorch enables accurate prediction across 100 skin shades. Simultaneously, we showcase the transformative potential of GANs in the realm of black hairstyles. By training GANs on a curated dataset of diverse hair textures and styles, we illustrate how PyTorch facilitates the generation of lifelike images that celebrate the beauty and diversity of black hair. Attendees will gain insights into the data preprocessing, model training, and evaluation processes and and learn how PyTorch empowers developers to build inclusive solutions.

Data-dependent shapes are ubiquitous whenever you want to take advantage of sparsity in your data representation, whether it is in recommendation systems, mixture of experts or other use cases. We have made a lot of improvements to torch.compile's support for capturing and compiling data dependent shapes, but they also require some user knowledge to work with effectively. This talk will give an overview of PT2's facilities for data dependent compute and how to use them effectively.

We will present vLLM, an open-source high-performance LLM inference engine built on top of PyTorch. Starting as a research project at UC Berkeley, vLLM has been one of the fastest and most popular LLM inference solutions in industry, reaching 20K+ stars and 350+ contributors. In this talk, we will cover how vLLM adopts various LLM inference optimizations and how it supports various AI accelerators such as AMD GPUs, Google TPUs, and AWS Inferentia. Also, we will discuss how vLLM benefits from PyTorch 2 and its ecosystem.

The PyTorch team has improved training speed by an order of magnitude for teams at Meta working on Small-to-Large-Scale MultiModal Video models. In this talk we’ll share our learnings on reducing GPU starvation by overcoming data loading challenges such as dealing with large distributed datasets, worker imbalance, compute-bottlenecks due to parallel video decoding and sampling, checkpointing, and debuggability. As part of our commitment to open-source, we are releasing a new decoding library and updating existing PyTorch libraries on GitHub, and invite feedback and contributions from the community.

torchtitan is a proof-of-concept for Large-scale LLM training using native PyTorch. It is a repo that showcases PyTorch's latest distributed training features in a clean, minimal codebase. We show-cased end to end large scale training features enablement: 1. 3D/4D Parallelism 2. Efficient distributed checkpoint save/load/resharding 3. Many efficient training techniques including Float8, torch.compile, activation checkpoint, etc.

Have you ever hit an OOM (and wished you had more VRAM)? Who hasn't! Hop on the bus with us and feel the road become smoother as we talk about stacking together techniques like FSDP2 + QLoRa + CPU Offloading + Fused ADAM (thanks Intel) + more in PyTorch native. We will give an overview of these techniques as well as the hard edges we solved in their composition. Curious for more? Or...still OOMing? We also plan on discussing our more researchy work on offloading, pagedness, and low precision optimizers.

In this session we will cover various ways teams at google are working to help the Pytorch community achieve performance and scale from cloud to edge. We will cover how Google Cloud customers can use PyTorch and OpenXLA to get competitive performance for their ML workloads.  We’ll also cover how Google AI Edge Torch works with Pytorch to help developers integrate LLMs, vision models and more to easily create new edge applications that can run on a wide set of devices.

In this talk, we will present the latest advancements in torch.compile for distributed training via DDP and FSDP. We will first introduce Compiled Autograd, a torch.compile mode to fully capture the backpropagation step, including the communication collective operators used in distributed. We will then cover the improvements this new approach brought to Compiled DDP/FSDP, notably by removing DDP/FSDP graph breaks which brings the potential of improving compute/communication overlap.

Understanding how to effectively size a production grade LLM deployment requires understanding of the model(s), the compute hardware, quantization and parallelization methods, KV Cache budgets, input and output token length predictions, model adapter management and much more. - Why LLM inference is different to standard deep learning inference - Current and future NVIDIA GPU overview - which GPU(s) for which models and why - Understanding the importance of building inference engines - Deep recap on the attention mechanism along with different types of popular attention mechanisms used in production - Deep dive on KV Cache and managing KV Cache budgets - Parallelism (reducing latency) - mainly tensor parallelism, but data, sequence, pipeline, and expert parallelism will be highlighted - Quantization methods on weights, activations, and KV Cache to reduce engine sizes for more effective GPU utilization - Increasing throughput with inflight batching and other techniques - Detailed performance analysis of LLM deployments looking at Time to first token, inter-token latencies, llm deployment characterizations, and more that can help reduce deployment costs

The integration of Intel GPU support into PyTorch marks a pivotal enhancement for PyTorch device and runtime. We generalized the PyTorch device and runtime to accommodate streaming devices. The generalization not only facilitates the deployment of PyTorch on ubiquitous hardware but also makes the integration of different HW backends easier. In addition, PyTorch with Intel GPU supports various Intel GPUs from the data center to the client. It enriches and democratizes PyTorch HW ecosystem. Particularly in AIPC scenarios where Intel's integrated and discrete GPUs are prevalent, Pytorch with Intel GPU can deliver promising performance and improved OOB experience in the AIPC domain that can extend PyTorch's applicability significantly.

In an era where artificial intelligence reigns supreme, the statistics are both perplexing and thought-provoking – only a mere 13% of large language models manage to transcend the realms of research and enter the practical world of production. Who bears the responsibility when these models err, spewing out biased or discriminatory outputs? It's time to demystify the complex landscape of machine learning ethics and carve a path towards a brighter, more accountable future! In this talk, firstly, we will navigate the profound impacts of large language models across diverse domains, from the lifesaving advances in medicine to safeguarding our nations through enhanced security protocols. Secondly, as we marvel at data-driven decisions laid by these models, we will confront the darker shadows cast by – the looming spectre of bias in the data. Finally, we will delve deep into the art of building interpretable models and navigating the maze of ethical considerations. Through a live demonstration in PyTorch, we will witness how to craft unbiased, transparent, and explainable models.

This presentation will dive into the development of the ONNXRuntime (ORT) backend for torch.compile. We'll cover the implementation process, starting with a PyTorch 2.0 generated FX graph, highlighting the unique challenges encountered when serving ORT-specific scenarios and how we solved them. Attendees will gain insights into optimizing performance, overcoming integration hurdles, and achieving efficient execution. Whether you're a developer looking to extend PyTorch's capabilities for your own use cases, keen to learn about ONNX Runtime, or interested in backend performance optimization, and the many steps we've taken to get to where we are now, this session promises valuable takeaways and practical knowledge.

Artificial Intelligence (AI) has the potential to revolutionize various sectors, including environmental conservation and water management. However, the deployment of AI technologies raises ethical questions about the environmental impact, particularly water resources. This presentation will discuss the ethical implications of AI concerning water while also exploring how AI can both positively and negatively affect water resources along with the broader ecosystem. My goal is to facilitate a critical conversation around how to balance technological advancements with environmental stewardship. Objectives: Understanding Ethical Implications: Provide an in depth overview of how AI impacts water resources. Focus on ethical concerns related to AI's water footprint, including, but not limited to energy consumption and water usage in data centers. Explore Positive Applications: Talk about the possible successful implementations of AI in water conservation, pollution monitoring, and efficient resource management. Discuss potential future applications where AI could contribute to sustainable water management and connect stakeholders to address ethical concerns and solutions.