TorchMetrics v0.11 — Multimodal and nominal We are happy to announce that Torchmetrics v0.11 is now publicly available. In Torchmetrics v0.11 we have primarily focused on the cleanup of the large classification refactor from v0.10 and adding new metrics. With v0.11 are crossing 90+ metrics in Torchmetrics nearing the milestone of having 100+ metrics. Classification refactor First, we would like to highlight that from v0.11 of Torchmetrics the classification refactor that was done in v0.11 is now fully implemented and has taken effect. You can read more about the refactor in this blog ..read more

Lightning Transformers 0.2 — New ?Tasks, Community Features, and Big Model Training & Inference Pairing ? Transformers and Lightning has become increasingly popular, leveraging Lightning to hide away the boilerplate of your training code, whilst training using the extensive models and datasets library that Transformers provides. Today we’re announcing Lightning Transformers 0.2 packed with new features, including the new Vision Transformers Image Classification Task, Sparsity support with SparseML, ? Hub Checkpoint Support, Big Transformers Inference, and DeepSpeed Training support ..read more

TorchMetrics v0.10 — Large changes to classifications TorchMetrics v0.10 is now out, significantly changing the whole classification package. This blog post will go over the reasons why the classification package needs to be refactored, what it means for our end users, and finally, what benefits it gives. A guide on upgrading your code to the recent changes can be found near the bottom. Why the classification metrics need to change We have long known that there were some underlying problems with how we initially structured the classification package. Essentially, classification tasks ..read more

Photo by Karolina GrabowskaTorchMetrics v0.9 — Faster forward TorchMetrics v0.9 is now out, and it brings significant changes to how the forward method works. This blog post goes over these improvements and how they affect both users of TorchMetrics and users that implement custom metrics. TorchMetrics v0.9 also includes several new metrics and bug fixes. TL;DR: If you only use metrics from TorchMetrics, you do not need to change anything in your code. If you implement your own metrics, you need to set one new class property. The Story of the Forward Method Since the beginning o ..read more

Authors: Sunil Srinivasa, Tian Lan, Huan Wang, Stephan Zheng, and Donald Rose Reinforcement Learning: Agents Learn by Maximizing Rewards Reinforcement Learning (RL) is a subfield of Machine Learning (ML) that deals with how intelligent agents should act in an environment when they wish to maximize a reward. This reward can be defined in various ways depending on the domain. The basic concept of RL is learning through interaction with a simulation or the real world, making minimal assumptions about how that simulation works. As such, RL is very flexible. For instance, RL can optimize ..read more

PyTorch Lightning 1.6: Support Intel’s Habana Accelerator, New efficient DDP strategy (Bagua), Manual Fault-tolerance, Stability and Reliability. PyTorch Lightning 1.6 is the work of 99 contributors who have worked on features, bug-fixes, and documentation for a total of over 750 commits since 1.5. This is our most active release yet. Here are some highlights: Introducing Intel’s Habana Accelerator Lightning 1.6 now supports the Habana® framework, which includes Gaudi® AI training processors. Their heterogeneous architecture includes a cluster of fully programmable Tensor Processing Cores (TPC ..read more

We recently added support for Habana’s Gaudi AI Processors, which can be used to accelerate deep learning training workloads. We also covered the benefits of using Habana and how to leverage the Habana Accelerator with PyTorch Lightning on a livestream: https://medium.com/media/43cb8606939db478e6bee1126227a000/hrefWhat is the Habana Gaudi AI Processor? ? Habana Gaudi was designed from the ground up to maximize training throughput and efficiency. ? The processors are built on a heterogeneous architecture with a cluster of fully programmable Tensor Processing Cores (TPC), along with its associa ..read more

PyTorch’s Meta Tensors can save you huge amounts of time. PyTorch Lightning, together with DeepSpeed and just a single line of code, allows you to train large, billion-parameter models even faster. PyTorch Lighting is a lightweight PyTorch wrapper for high-performance AI research. PyTorch Lightning provides true flexibility by reducing the engineering boilerplate and resources required to implement state-of-the-art AI. Organizing PyTorch code with Lightning enables seamless training on multiple GPUs, TPUs, and CPUs and the use of difficult to implement best practices such as model shardi ..read more

Illustration from Pixabay.TorchMetrics v0.8 — Paper, Faster collection, and more metrics! We are excited to announce that TorchMetrics v0.8 is now available. This release includes several new metrics in the classification and image domains as well as performance improvements for those working with metrics collections. TorchMetrics paper Since the last version, we wrote a short paper about TorchMetrics that was published in the Journal of Open Source Software (JOSS). You can download it here. If you’re using TorchMetrics in your research, we would appreciate it if you cited the paper ..read more

Bagua: A new, efficient distributed training strategy available in PyTorch Lightning We’ve added support for a new distributed training strategy in collaboration with the Bagua team. We also covered the benefits of this training strategy in a live stream: https://medium.com/media/5990fb615236caecef6b734fbbbd33ab/hrefWhat is Bagua? BaguaSys/Bagua is a deep learning acceleration framework for PyTorch developed by AI platform@Kuaishou Technology and DS3 Lab@ETH. Bagua supports multiple, advanced distributed training algorithms with state-of-the-art system relaxation techniques ..read more