MobileQuant: Mobile-friendly Quantization for On-device Language Models

Large language models (LLMs) have revolutionized language processing, delivering outstanding results across multiple applications. However, deploying LLMs on edge devices poses several challenges with respect to memory, energy, and compute costs, limiting their widespread use in devices such as mobile phones. A promising solution is to reduce the number of bits used to represent weights and activations. While existing works have found partial success at quantizing LLMs to lower bitwidths, e.g. 4-bit weights, quantizing activations beyond 16 bits often leads to large computational overheads due to poor on-device quantization support, or a considerable accuracy drop. Yet, 8-bit activations are very attractive for on-device deployment as they would enable LLMs to fully exploit mobile-friendly hardware, e.g. Neural Processing Units (NPUs). In this work, we make a first attempt to facilitate the on-device deployment of LLMs using integer-only quantization. We first investigate the limitations of existing quantization methods for on-device deployment, with a special focus on activation quantization. We then address these limitations by introducing a simple post-training quantization method, named MobileQuant, that extends previous weight equivalent transformation works by jointly optimizing the weight transformation and activation range parameters in an end-to-end manner. MobileQuant demonstrates superior capabilities over existing methods by 1) achieving near-lossless quantization on a wide range of LLM benchmarks, 2) reducing latency and energy consumption by 20\%-50\% compared to current on-device quantization strategies, 3) requiring limited compute budget, 4) being compatible with mobile-friendly compute units, e.g. NPU.

Linear-Complexity Self-Supervised Learning for Speech Processing

Self-supervised learning (SSL) models usually require weeks of pre-training with dozens of high-end GPUs. These models typically have a multi-headed self-attention (MHSA) context encoder. However, MHSA takes quadratic time and space in the input length, contributing to the high pre-training cost. Linear-complexity alternatives to MHSA have been proposed. For instance, in supervised training, the SummaryMixing model is the first to outperform MHSA across multiple speech processing tasks. However, these cheaper alternatives have not been explored for SSL yet. This paper studies a linear-complexity context encoder for SSL for the first time. With better or equivalent performance for the downstream tasks of the MP3S benchmark, SummaryMixing reduces the pre-training time and peak VRAM of wav2vec 2.0 model by 18% and by 23%, respectively, leading to the pre-training of a 155M wav2vec 2.0 model finished within one week with 4 Tesla A100 GPUs. Code is available at https://github.com/SamsungLabs/SummaryMixing.

Fast Inference Through The Reuse Of Attention Maps In Diffusion Models

Text-to-image diffusion models have demonstrated unprecedented abilities at flexible and realistic image synthesis. However, the iterative process required to produce a single image is costly and incurs a high latency, prompting researchers to further investigate its efficiency. Typically, improvements in latency have been achieved in two ways: (1) training smaller models through knowledge distillation (KD); and (2) adopting techniques from ODE-theory to facilitate larger step sizes. In contrast, we propose a training-free approach that does not alter the step-size of the sampler. Specifically, we find the repeated calculation of attention maps to be both costly and redundant; therefore, we propose a structured reuse of attention maps during sampling. Our initial reuse policy is motivated by rudimentary ODE-theory, which suggests that reuse is most suitable late in the sampling procedure. After noting a number of limitations in this theoretical approach, we empirically search for a better policy. Unlike methods that rely on KD, our reuse policies can easily be adapted to a variety of setups in a plug-and-play manner. Furthermore, when applied to Stable Diffusion-1.5, our reuse policies reduce latency with minimal repercussions on sample quality.

Sumformer: A Linear-Complexity Alternative to Self-Attention for Speech Recognition

Modern speech recognition systems rely on self-attention. Unfortunately, token mixing with self-attention takes quadratic time in the length of the speech utterance, slowing down inference as well as training and increasing memory consumption. Cheaper alternatives to self-attention for ASR have been developed, but fail to consistently reach the same level of accuracy. In practice, however, the self-attention weights of trained speech recognizers take the form of a global average over time. This paper, therefore, proposes a linear-time alternative to self-attention for speech recognition. It summarises a whole utterance with the mean over vectors for all time steps. This single summary is then combined with time-specific information. We call this method ``Summary Mixing''. Introducing Summary Mixing in state-of-the-art ASR models makes it feasible to preserve or exceed previous speech recognition performance while lowering the training and inference times by up to 27% and reducing the memory budget by a factor of two.

Cross-Attention is all you need: Real-Time Streaming Transformers for Personalised Speech Enhancement

Personalised speech enhancement (PSE), which extracts only the speech of a target user and removes everything else from a recorded audio clip, can potentially improve users' experiences of audio AI modules deployed in the wild. To support a large variety of downstream audio tasks, such as real-time ASR and audio-call enhancement, a PSE solution should operate in a streaming mode, i.e., input audio cleaning should happen in real-time with a small latency and real-time factor. Personalisation is typically achieved by extracting a target speaker's voice profile from an enrolment audio, in the form of a static embedding vector, and then using it to condition the output of a PSE model. However, a fixed target speaker embedding may not be optimal under all conditions. In this work, we present a streaming Transformer-based PSE model and propose a novel cross-attention approach that gives adaptive target speaker representations. We present extensive experiments and show that our proposed cross-attention approach outperforms competitive baselines consistently, even when our model is only approximately half the size.

Defensive Tensorization

We propose defensive tensorization, an adversarial defence technique that leverages a latent high-order factorization of the network. The layers of a network are first expressed as factorized tensor layers. Tensor dropout is then applied in the latent subspace, therefore resulting in dense reconstructed weights, without the sparsity or perturbations typically induced by the randomization.Our approach can be readily integrated with any arbitrary neural architecture and combined with techniques like adversarial training. We empirically demonstrate the effectiveness of our approach on standard image classification benchmarks. We validate the versatility of our approach across domains and low-precision architectures by considering an audio classification task and binary networks. In all cases, we demonstrate improved performance compared to prior works.

Bunched LPCNet : Vocoder for Low-cost Neural Text-To-Speech Systems

LPCNet is an efficient vocoder that combines linear prediction and deep neural network modules to keep the computational complexity low. In this work, we present two techniques to further reduce it's complexity, aiming for a low-cost LPCNet vocoder-based neural Text-to-Speech (TTS) System. These techniques are: 1) Sample-bunching, which allows LPCNet to generate more than one audio sample per inference; and 2) Bit-bunching, which reduces the computations in the final layer of LPCNet. With the proposed bunching techniques, LPCNet, in conjunction with a Deep Convolutional TTS (DCTTS) acoustic model, shows a 2.19x improvement over the baseline run-time when running on a mobile device, with a less than 0.1 decrease in TTS mean opinion score (MOS).

Iterative Compression of End-to-End ASR Model using AutoML

Increasing demand for on-device Automatic Speech Recognition (ASR) systems has resulted in renewed interests in developing automatic model compression techniques. Past research have shown that AutoML-based Low Rank Factorization (LRF) technique, when applied to an end-to-end Encoder-Attention-Decoder style ASR model, can achieve a speedup of up to 3.7x, outperforming laborious manual rank-selection approaches. However, we show that current AutoML-based search techniques only work up to a certain compression level, beyond which they fail to produce compressed models with acceptable word error rates (WER). In this work, we propose an iterative AutoML-based LRF approach that achieves over 5x compression without degrading the WER, thereby advancing the state-of-the-art in ASR compression.

MobiSR: Efficient On-Device Super-Resolution through Heterogeneous Mobile Processors

In recent years, convolutional networks have demonstrated unprecedented performance in the image restoration task of super-resolution (SR). SR entails the upscaling of a single low-resolution image in order to meet application-specific image quality demands and plays a key role in mobile devices. To comply with privacy regulations and reduce the overhead of cloud computing, executing SR models locally on-device constitutes a key alternative approach. Nevertheless, the excessive compute and memory requirements of SR workloads pose a challenge in mapping SR networks on resource-constrained mobile platforms. This work presents MobiSR, a novel framework for performing efficient super-resolution on-device. Given a target mobile platform, the proposed framework considers popular model compression techniques and traverses the design space to reach the highest performing trade-off between image quality and processing speed. At run time, a novel scheduler dispatches incoming image patches to the appropriate model-engine pair based on the patch's estimated upscaling difficulty in order to meet the required image quality with minimum processing latency. Quantitative evaluation shows that the proposed framework yields on-device SR designs that achieve an average speedup of 2.13x over highly-optimized parallel difficulty-unaware mappings and 4.79x over highly-optimized single compute engine implementations.

Cross-modal Recurrent Models for Weight Objective Prediction from Multimodal Time-series Data

We analyse multimodal time-series data corresponding to weight, sleep and steps measurements. We focus on predicting whether a user will successfully achieve his/her weight objective. For this, we design several deep long short-term memory (LSTM) architectures, including a novel cross-modal LSTM (X-LSTM), and demonstrate their superiority over baseline approaches. The X-LSTM improves parameter efficiency by processing each modality separately and allowing for information flow between them by way of recurrent cross-connections. We present a general hyperparameter optimisation technique for X-LSTMs, which allows us to significantly improve on the LSTM and a prior state-of-the-art cross-modal approach, using a comparable number of parameters. Finally, we visualise the model's predictions, revealing implications about latent variables in this task.