We introduce MuJoCo Playground, a fully open-source framework for robot learning built with MJX, with the express goal of streamlining simulation, training, and sim-to-real transfer onto robots. With a simple "pip install playground", researchers can train policies in minutes on a single GPU. Playground supports diverse robotic platforms, including quadrupeds, humanoids, dexterous hands, and robotic arms, enabling zero-shot sim-to-real transfer from both state and pixel inputs. This is achieved through an integrated stack comprising a physics engine, batch renderer, and training environments. Along with video results, the entire framework is freely available at playground.mujoco.org

Recently, ChatGPT or InstructGPT like large language models (LLM) has made a significant impact in the AI world. Many works have attempted to reproduce the complex InstructGPT's training pipeline, namely Reinforcement Learning with Human Feedback (RLHF). However, the mainstream distributed RLHF training methods typically adopt a fixed model placement strategy, referred to as the Flattening strategy. This strategy treats all four interdependent models involved in RLHF as a single entity, distributing them across all devices and applying parallelism techniques designed for a single model, regardless of the different workloads inherent to each model. As a result, this strategy exacerbates the generation bottlenecks in the RLHF training and degrades the overall training efficiency. To address these issues, we propose an adaptive model placement framework that offers two flexible model placement strategies. The Interleaving strategy helps reduce memory redundancy and communication costs of RLHF training by placing models without dependencies on exclusive devices with careful orchestration. On the other hand, the Separation strategy improves the throughput of model training by separating the training and inference runtime of the RLHF pipeline with additional shadow models. Furthermore, our framework provides a simple user interface and allows for the agile allocation of models across devices in a fine-grained manner for various training scenarios, involving models of varying sizes and devices of different scales. Extensive experiments have demonstrated that our Interleaving and Separation strategies can achieve notable improvements up to 11X, compared to the current SOTA approaches. The results highlight the effectiveness and adaptability of our approaches in accelerating the training of distributed RLHF.

Large language models (LLMs) such as GPT-3, OPT, and LLaMA have demonstrated remarkable accuracy in a wide range of tasks. However, training these models can incur significant expenses, often requiring tens of thousands of GPUs for months of continuous operation. Typically, this training is carried out in specialized GPU clusters equipped with homogeneous high-speed Remote Direct Memory Access (RDMA) network interface cards (NICs). The acquisition and maintenance of such dedicated clusters is challenging. Current LLM training frameworks, like Megatron-LM and Megatron-DeepSpeed, focus primarily on optimizing training within homogeneous cluster settings. In this paper, we introduce Holmes, a training framework for LLMs that employs thoughtfully crafted data and model parallelism strategies over the heterogeneous NIC environment. Our primary technical contribution lies in a novel scheduling method that intelligently allocates distinct computational tasklets in LLM training to specific groups of GPU devices based on the characteristics of their connected NICs. Furthermore, our proposed framework, utilizing pipeline parallel techniques, demonstrates scalability to multiple GPU clusters, even in scenarios without high-speed interconnects between nodes in distinct clusters. We conducted comprehensive experiments that involved various scenarios in the heterogeneous NIC environment. In most cases, our framework achieves performance levels close to those achievable with homogeneous RDMA-capable networks (InfiniBand or RoCE), significantly exceeding training efficiency within the pure Ethernet environment. Additionally, we verified that our framework outperforms other mainstream LLM frameworks under heterogeneous NIC environment in terms of training efficiency and can be seamlessly integrated with them.

It is widely acknowledged that large models have the potential to deliver superior performance across a broad range of domains. Despite the remarkable progress made in the field of machine learning systems research, which has enabled the development and exploration of large models, such abilities remain confined to a small group of advanced users and industry leaders, resulting in an implicit technical barrier for the wider community to access and leverage these technologies. In this paper, we introduce PyTorch Fully Sharded Data Parallel (FSDP) as an industry-grade solution for large model training. FSDP has been closely co-designed with several key PyTorch core components including Tensor implementation, dispatcher system, and CUDA memory caching allocator, to provide non-intrusive user experiences and high training efficiency. Additionally, FSDP natively incorporates a range of techniques and settings to optimize resource utilization across a variety of hardware configurations. The experimental results demonstrate that FSDP is capable of achieving comparable performance to Distributed Data Parallel while providing support for significantly larger models with near-linear scalability in terms of TFLOPS.

Neural processes are a family of probabilistic models that inherit the flexibility of neural networks to parameterize stochastic processes. Despite providing well-calibrated predictions, especially in regression problems, and quick adaptation to new tasks, the Gaussian assumption that is commonly used to represent the predictive likelihood fails to capture more complicated distributions such as multimodal ones. To overcome this limitation, we propose Conditional Quantile Neural Processes (CQNPs), a new member of the neural processes family, which exploits the attractive properties of quantile regression in modeling the distributions irrespective of their form. By introducing an extension of quantile regression where the model learns to focus on estimating informative quantiles, we show that the sampling efficiency and prediction accuracy can be further enhanced. Our experiments with real and synthetic datasets demonstrate substantial improvements in predictive performance compared to the baselines, and better modeling of heterogeneous distributions' characteristics such as multimodality.

I am one of the contributors to the Spark NLP open-source project and just recently this library started supporting end-to-end Vision Transformers (ViT) models. I use Spark NLP and other ML/DL open-source libraries for work daily and I have decided to deploy a ViT pipeline for a state-of-the-art image classification task and provide in-depth comparisons between Hugging Face and Spark NLP. The purpose of this article is to demonstrate how to scale out Vision Transformer (ViT) models from Hugging Face and deploy them in production-ready environments for accelerated and high-performance inference. By the end, we will scale a ViT model from Hugging Face by 25x times (2300%) by using Databricks, Nvidia, and Spark NLP. Back in 2017, a group of researchers at Google AI published a paper that introduced a transformer model architecture that changed all Natural Language Processing (NLP) standards.

Data scientists and machine learning developers are building and training very large models these days with more extensive GPU memory needs. Many of these larger ML applications need more than one NVIDIA GPU device on the vSphere servers on which they operate or they may need to communicate between separate GPUs over the local network. This can be done for the purpose of expanding the overall GPU framebuffer memory capacity or for other reasons. Servers now exist on the market with eight or more physical GPUs in them and that number of GPUs per server will likely grow over time. With vSphere 8, you have the capability to add up to 8 virtual GPUs (vGPUs) to one VM.

In this paper, we provide a fine-grain machine learning-based method, PerfNetV2, which improves the accuracy of our previous work for modeling the neural network performance on a variety of GPU accelerators. Given an application, the proposed method can be used to predict the inference time and training time of the convolutional neural networks used in the application, which enables the system developer to optimize the performance by choosing the neural networks and/or incorporating the hardware accelerators to deliver satisfactory results in time. Furthermore, the proposed method is capable of predicting the performance of an unseen or non-existing device, e.g. a new GPU which has a higher operating frequency with less processor cores, but more memory capacity. This allows a system developer to quickly search the hardware design space and/or fine-tune the system configuration. Compared to the previous works, PerfNetV2 delivers more accurate results by modeling detailed host-accelerator interactions in executing the full neural networks and improving the architecture of the machine learning model used in the predictor. Our case studies show that PerfNetV2 yields a mean absolute percentage error within 13.1% on LeNet, AlexNet, and VGG16 on NVIDIA GTX-1080Ti, while the error rate on a previous work published in ICBD 2018 could be as large as 200%.