FLUX: Fast Software-based Communication Overlap On GPUs Through Kernel Fusion

Large deep learning models have demonstrated strong ability to solve many tasks across a wide range of applications. Those large models typically require training and inference to be distributed. Tensor parallelism is a common technique partitioning computation of an operation or layer across devices to overcome the memory capacity limitation of a single processor, and/or to accelerate computation to meet a certain latency requirement. However, this kind of parallelism introduces additional communication that might contribute a significant portion of overall runtime. Thus limits scalability of this technique within a group of devices with high speed interconnects, such as GPUs with NVLinks in a node. This paper proposes a novel method, Flux, to significantly hide communication latencies with dependent computations for GPUs. Flux over-decomposes communication and computation operations into much finer-grained operations and further fuses them into a larger kernel to effectively hide communication without compromising kernel efficiency. Flux can potentially overlap up to 96% of communication given a fused kernel. Overall, it can achieve up to 1.24x speedups for training over Megatron-LM on a cluster of 128 GPUs with various GPU generations and interconnects, and up to 1.66x and 1.30x speedups for prefill and decoding inference over vLLM on a cluster with 8 GPUs with various GPU generations and interconnects.

MegaScale: Scaling Large Language Model Training to More Than 10,000 GPUs

We present the design, implementation and engineering experience in building and deploying MegaScale, a production system for training large language models (LLMs) at the scale of more than 10,000 GPUs. Training LLMs at this scale brings unprecedented challenges to training efficiency and stability. We take a full-stack approach that co-designs the algorithmic and system components across model block and optimizer design, computation and communication overlapping, operator optimization, data pipeline, and network performance tuning. Maintaining high efficiency throughout the training process (i.e., stability) is an important consideration in production given the long extent of LLM training jobs. Many hard stability issues only emerge at large scale, and in-depth observability is the key to address them. We develop a set of diagnosis tools to monitor system components and events deep in the stack, identify root causes, and derive effective techniques to achieve fault tolerance and mitigate stragglers. MegaScale achieves 55.2% Model FLOPs Utilization (MFU) when training a 175B LLM model on 12,288 GPUs, improving the MFU by 1.34x compared to Megatron-LM. We share our operational experience in identifying and fixing failures and stragglers. We hope by articulating the problems and sharing our experience from a systems perspective, this work can inspire future LLM systems research.

Relax: Composable Abstractions for End-to-End Dynamic Machine Learning

Dynamic shape computations have become critical in modern machine learning workloads, especially in emerging large language models. The success of these models has driven demand for deploying them to a diverse set of backend environments. In this paper, we present Relax, a compiler abstraction for optimizing end-to-end dynamic machine learning workloads. Relax introduces first-class symbolic shape annotations to track dynamic shape computations globally across the program. It also introduces a cross-level abstraction that encapsulates computational graphs, loop-level tensor programs, and library calls in a single representation to enable cross-level optimizations. We build an end-to-end compilation framework using the proposed approach to optimize dynamic shape models. Experimental results on large language models show that Relax delivers performance competitive with state-of-the-art hand-optimized systems across platforms and enables deployment of emerging dynamic models to a broader set of environments, including mobile phones, embedded devices, and web browsers.

Federated Remote Physiological Measurement with Imperfect Data

The growing need for technology that supports remote healthcare is being acutely highlighted by an aging population and the COVID-19 pandemic. In health-related machine learning applications the ability to learn predictive models without data leaving a private device is attractive, especially when these data might contain features (e.g., photographs or videos of the body) that make identifying a subject trivial and/or the training data volume is large (e.g., uncompressed video). Camera-based remote physiological sensing facilitates scalable and low-cost measurement, but is a prime example of a task that involves analysing high bit-rate videos containing identifiable images and sensitive health information. Federated learning enables privacy-preserving decentralized training which has several properties beneficial for camera-based sensing. We develop the first mobile federated learning camera-based sensing system and show that it can perform competitively with traditional state-of-the-art supervised approaches. However, in the presence of corrupted data (e.g., video or label noise) from a few devices the performance of weight averaging quickly degrades. To address this, we leverage knowledge about the expected noise profile within the video to intelligently adjust how the model weights are averaged on the server. Our results show that this significantly improves upon the robustness of models even when the signal-to-noise ratio is low

EfficientPhys: Enabling Simple, Fast and Accurate Camera-Based Vitals Measurement

Camera-based physiological measurement is a growing field with neural models providing state-the-art-performance. Prior research have explored various ``end-to-end'' models; however these methods still require several preprocessing steps. These additional operations are often non-trivial to implement making replication and deployment difficult and can even have a higher computational budget than the ``core'' network itself. In this paper, we propose two novel and efficient neural models for camera-based physiological measurement called EfficientPhys that remove the need for face detection, segmentation, normalization, color space transformation or any other preprocessing steps. Using an input of raw video frames, our models achieve state-of-the-art accuracy on three public datasets. We show that this is the case whether using a transformer or convolutional backbone. We further evaluate the latency of the proposed networks and show that our most light weight network also achieves a 33% improvement in efficiency.

Automated Backend-Aware Post-Training Quantization

Quantization is a key technique to reduce the resource requirement and improve the performance of neural network deployment. However, different hardware backends such as x86 CPU, NVIDIA GPU, ARM CPU, and accelerators may demand different implementations for quantized networks. This diversity calls for specialized post-training quantization pipelines to built for each hardware target, an engineering effort that is often too large for developers to keep up with. We tackle this problem with an automated post-training quantization framework called HAGO. HAGO provides a set of general quantization graph transformations based on a user-defined hardware specification and implements a search mechanism to find the optimal quantization strategy while satisfying hardware constraints for any model. We observe that HAGO achieves speedups of 2.09x, 1.97x, and 2.48x on Intel Xeon Cascade Lake CPUs, NVIDIA Tesla T4 GPUs, ARM Cortex-A CPUs on Raspberry Pi4 relative to full precision respectively, while maintaining the highest reported post-training quantization accuracy in each case.

SplitSR: An End-to-End Approach to Super-Resolution on Mobile Devices

Super-resolution (SR) is a coveted image processing technique for mobile apps ranging from the basic camera apps to mobile health. Existing SR algorithms rely on deep learning models with significant memory requirements, so they have yet to be deployed on mobile devices and instead operate in the cloud to achieve feasible inference time. This shortcoming prevents existing SR methods from being used in applications that require near real-time latency. In this work, we demonstrate state-of-the-art latency and accuracy for on-device super-resolution using a novel hybrid architecture called SplitSR and a novel lightweight residual block called SplitSRBlock. The SplitSRBlock supports channel-splitting, allowing the residual blocks to retain spatial information while reducing the computation in the channel dimension. SplitSR has a hybrid design consisting of standard convolutional blocks and lightweight residual blocks, allowing people to tune SplitSR for their computational budget. We evaluate our system on a low-end ARM CPU, demonstrating both higher accuracy and up to 5 times faster inference than previous approaches. We then deploy our model onto a smartphone in an app called ZoomSR to demonstrate the first-ever instance of on-device, deep learning-based SR. We conducted a user study with 15 participants to have them assess the perceived quality of images that were post-processed by SplitSR. Relative to bilinear interpolation -- the existing standard for on-device SR -- participants showed a statistically significant preference when looking at both images (Z=-9.270, p<0.01) and text (Z=-6.486, p<0.01).

MetaPhys: Unsupervised Few-Shot Adaptation for Non-Contact Physiological Measurement

There are large individual differences in physiological processes, making designing personalized health sensing algorithms challenging. Existing machine learning systems struggle to generalize well to unseen subjects or contexts, especially in video-based physiological measurement. Although fine-tuning for a user might address this issue, it is difficult to collect large sets of training data for specific individuals because supervised algorithms require medical-grade sensors for generating the training target. Therefore, learning personalized or customized models from a small number of unlabeled samples is very attractive as it would allow fast calibrations. In this paper, we present a novel unsupervised meta-learning approach called MetaPhys for learning personalized cardiac signals from 18-seconds of unlabeled video data. We evaluate our proposed approach on two benchmark datasets and demonstrate superior performance in cross-dataset evaluation with substantial reductions (42% to 44%) in errors compared with state-of-the-art approaches. Visualization of attention maps and ablation experiments reveal how the model adapts to each subject and why our proposed approach leads to these improvements. We have also demonstrated our proposed method significantly helps reduce the bias in skin type.

Exploring the Memorization-Generalization Continuum in Deep Learning

Human learners appreciate that some facts demand memorization whereas other facts support generalization. For example, English verbs have irregular cases that must be memorized (e.g., go->went) and regular cases that generalize well (e.g., kiss->kissed, miss->missed). Likewise, deep neural networks have the capacity to memorize rare or irregular forms but nonetheless generalize across instances that share common patterns or structures. We analyze how individual instances are treated by a model on the memorization-generalization continuum via a consistency score. The score is the expected accuracy of a particular architecture for a held-out instance on a training set of a fixed size sampled from the data distribution. We obtain empirical estimates of this score for individual instances in multiple datasets, and we show that the score identifies out-of-distribution and mislabeled examples at one end of the continuum and regular examples at the other end. We explore three proxies to the consistency score: kernel density estimation on input and hidden representations; and the time course of training, i.e., learning speed. In addition to helping to understand the memorization versus generalization dynamics during training, the C-score proxies have potential application for out-of-distribution detection, curriculum learning, and active data collection.

Relay: A High-Level IR for Deep Learning

Frameworks for writing, compiling, and optimizing deep learning (DL) models have recently enabled progress in areas like computer vision and natural language processing. Extending these frameworks to accommodate the rapidly diversifying landscape of DL models and hardware platforms presents challenging tradeoffs between expressiveness, composability, and portability. We present Relay, a new intermediate representation (IR) and compiler framework for DL models. The functional, statically-typed Relay IR unifies and generalizes existing DL IRs and can express state-of-the-art models. Relay's expressive IR required careful design of the type system, automatic differentiation, and optimizations. Relay's extensible compiler can eliminate abstraction overhead and target new hardware platforms. The design insights from Relay can be applied to existing frameworks to develop IRs that support extension without compromising on expressivity, composibility, and portability. Our evaluation demonstrates that the Relay prototype can already provide competitive performance for a broad class of models running on CPUs, GPUs, and FPGAs.