LRQ: Optimizing Post-Training Quantization for Large Language Models by Learning Low-Rank Weight-Scaling Matrices

With the commercialization of large language models (LLMs), weight-activation quantization has emerged to compress and accelerate LLMs, achieving high throughput while reducing inference costs. However, existing post-training quantization (PTQ) techniques for quantizing weights and activations of LLMs still suffer from non-negligible accuracy drops, especially on massive multitask language understanding. To address this issue, we propose Low-Rank Quantization (LRQ) $-$ a simple yet effective post-training weight quantization method for LLMs that reconstructs the outputs of an intermediate Transformer block by leveraging low-rank weight-scaling matrices, replacing the conventional full weight-scaling matrices that entail as many learnable scales as their associated weights. Thanks to parameter sharing via low-rank structure, LRQ only needs to learn significantly fewer parameters while enabling the individual scaling of weights, thus boosting the generalization capability of quantized LLMs. We show the superiority of LRQ over prior LLM PTQ works under (i) $8$-bit weight and per-tensor activation quantization, (ii) $4$-bit weight and $8$-bit per-token activation quantization, and (iii) low-bit weight-only quantization schemes. Our code is available at \url{https://github.com/onliwad101/FlexRound_LRQ} to inspire LLM researchers and engineers.

Improving Multi-hop Logical Reasoning in Knowledge Graphs with Context-Aware Query Representation Learning

Multi-hop logical reasoning on knowledge graphs is a pivotal task in natural language processing, with numerous approaches aiming to answer First-Order Logic (FOL) queries. Recent geometry (e.g., box, cone) and probability (e.g., beta distribution)-based methodologies have effectively addressed complex FOL queries. However, a common challenge across these methods lies in determining accurate geometric bounds or probability parameters for these queries. The challenge arises because existing methods rely on linear sequential operations within their computation graphs, overlooking the logical structure of the query and the relation-induced information that can be gleaned from the relations of the query, which we call the context of the query. To address the problem, we propose a model-agnostic methodology that enhances the effectiveness of existing multi-hop logical reasoning approaches by fully integrating the context of the FOL query graph. Our approach distinctively discerns (1) the structural context inherent to the query structure and (2) the relation-induced context unique to each node in the query graph as delineated in the corresponding knowledge graph. This dual-context paradigm helps nodes within a query graph attain refined internal representations throughout the multi-hop reasoning steps. Through experiments on two datasets, our method consistently enhances the three multi-hop reasoning foundation models, achieving performance improvements of up to 19.5%. Our code is available at https://github.com/kjh9503/caqr.

HyperCLOVA X Technical Report

We introduce HyperCLOVA X, a family of large language models (LLMs) tailored to the Korean language and culture, along with competitive capabilities in English, math, and coding. HyperCLOVA X was trained on a balanced mix of Korean, English, and code data, followed by instruction-tuning with high-quality human-annotated datasets while abiding by strict safety guidelines reflecting our commitment to responsible AI. The model is evaluated across various benchmarks, including comprehensive reasoning, knowledge, commonsense, factuality, coding, math, chatting, instruction-following, and harmlessness, in both Korean and English. HyperCLOVA X exhibits strong reasoning capabilities in Korean backed by a deep understanding of the language and cultural nuances. Further analysis of the inherent bilingual nature and its extension to multilingualism highlights the model's cross-lingual proficiency and strong generalization ability to untargeted languages, including machine translation between several language pairs and cross-lingual inference tasks. We believe that HyperCLOVA X can provide helpful guidance for regions or countries in developing their sovereign LLMs.

Domain Generalization in LiDAR Semantic Segmentation Leveraged by Density Discriminative Feature Embedding

While significant progress has been achieved in LiDAR-based perception, domain generalization continues to present challenges, often resulting in reduced performance when encountering unfamiliar datasets due to domain discrepancies. One of the primary hurdles stems from the variability of LiDAR sensors, leading to inconsistencies in point cloud density distribution. Such inconsistencies can undermine the effectiveness of perception models. We address this challenge by introducing a new approach that acknowledges a fundamental characteristic of LiDAR: the variation in point density due to the distance from the LiDAR to the scene, and the number of beams relative to the field of view. Understanding this, we view each LiDAR's point cloud at various distances as having distinct density distributions, which can be consistent across different LiDAR models. With this insight, we propose the Density Discriminative Feature Embedding (DDFE) module, crafted to specifically extract features related to density while ensuring domain invariance across different LiDAR sensors. In addition, we introduce a straightforward but effective density augmentation technique, designed to broaden the density spectrum and enhance the capabilities of the DDFE. The proposed DDFE stands out as a versatile and lightweight domain generalization module. It can be seamlessly integrated into various 3D backbone networks, consistently outperforming existing state-of-the-art domain generalization approaches. We commit to releasing the source code publicly to foster community collaboration and advancement.

Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models

Large Language Models (LLMs) have recently demonstrated a remarkable success across various tasks. However, efficiently serving LLMs has been a challenge due to its large memory bottleneck, specifically in small batch inference settings (e.g. mobile devices). Weight-only quantization can be a promising approach, but sub-4 bit quantization remains a challenge due to large-magnitude activation outliers. To mitigate the undesirable outlier effect, we first propose per-IC quantization, a simple yet effective method that creates quantization groups within each input channel (IC) rather than the conventional per-output channel (OC). Our method is motivated by the observation that activation outliers affect the input dimension of the weight matrix, so similarly grouping the weights in the IC direction can isolate outliers to be within a group. We also find that activation outliers do not dictate quantization difficulty, and inherent weight sensitivities also exist. With per-IC quantization as a new outlier-friendly scheme, we then propose Adaptive Dimensions (AdaDim), a versatile quantization framework that can adapt to various weight sensitivity patterns. We demonstrate the effectiveness of AdaDim by augmenting prior methods such as Round-To-Nearest and GPTQ, showing significant improvements across various language modeling benchmarks for both base (up to +4.7% on MMLU) and instruction-tuned (up to +10% on HumanEval) LLMs.

FlexRound: Learnable Rounding based on Element-wise Division for Post-Training Quantization

Post-training quantization (PTQ) has been gaining popularity for the deployment of deep neural networks on resource-limited devices since unlike quantization-aware training, neither a full training dataset nor end-to-end training is required at all. As PTQ schemes based on reconstructing each layer or block output turn out to be effective to enhance quantized model performance, recent works have developed algorithms to devise and learn a new weight-rounding scheme so as to better reconstruct each layer or block output. In this work, we propose a simple yet effective new weight-rounding mechanism for PTQ, coined FlexRound, based on element-wise division instead of typical element-wise addition such that FlexRound enables jointly learning a common quantization grid size as well as a different scale for each pre-trained weight. Thanks to the reciprocal rule of derivatives induced by element-wise division, FlexRound is inherently able to exploit pre-trained weights when updating their corresponding scales, and thus, flexibly quantize pre-trained weights depending on their magnitudes. We empirically validate the efficacy of FlexRound on a wide range of models and tasks. To the best of our knowledge, our work is the first to carry out comprehensive experiments on not only image classification and natural language understanding but also natural language generation, assuming a per-tensor uniform PTQ setting. Moreover, we demonstrate, for the first time, that large language models can be efficiently quantized, with only a negligible impact on performance compared to half-precision baselines, achieved by reconstructing the output in a block-by-block manner.

Memory-Efficient Fine-Tuning of Compressed Large Language Models via sub-4-bit Integer Quantization

Parameter-efficient fine-tuning (PEFT) methods have emerged to mitigate the prohibitive cost of full fine-tuning large language models (LLMs). Nonetheless, the enormous size of LLMs impedes routine deployment. To address the issue, we present Parameter-Efficient and Quantization-aware Adaptation (PEQA), a novel quantization-aware PEFT technique that facilitates model compression and accelerates inference. PEQA operates through a dual-stage process: initially, the parameter matrix of each fully-connected layer undergoes quantization into a matrix of low-bit integers and a scalar vector; subsequently, fine-tuning occurs on the scalar vector for each downstream task. Such a strategy compresses the size of the model considerably, leading to a lower inference latency upon deployment and a reduction in the overall memory required. At the same time, fast fine-tuning and efficient task switching becomes possible. In this way, PEQA offers the benefits of quantization, while inheriting the advantages of PEFT. We compare PEQA with competitive baselines in comprehensive experiments ranging from natural language understanding to generation benchmarks. This is done using large language models of up to $65$ billion parameters, demonstrating PEQA's scalability, task-specific adaptation performance, and ability to follow instructions, even in extremely low-bit settings.

AlphaTuning: Quantization-Aware Parameter-Efficient Adaptation of Large-Scale Pre-Trained Language Models

There are growing interests in adapting large-scale language models using parameter-efficient fine-tuning methods. However, accelerating the model itself and achieving better inference efficiency through model compression has not been thoroughly explored yet. Model compression could provide the benefits of reducing memory footprints, enabling low-precision computations, and ultimately achieving cost-effective inference. To combine parameter-efficient adaptation and model compression, we propose AlphaTuning consisting of post-training quantization of the pre-trained language model and fine-tuning only some parts of quantized parameters for a target task. Specifically, AlphaTuning works by employing binary-coding quantization, which factorizes the full-precision parameters into binary parameters and a separate set of scaling factors. During the adaptation phase, the binary values are frozen for all tasks, while the scaling factors are fine-tuned for the downstream task. We demonstrate that AlphaTuning, when applied to GPT-2 and OPT, performs competitively with full fine-tuning on a variety of downstream tasks while achieving >10x compression ratio under 4-bit quantization and >1,000x reduction in the number of trainable parameters.

CogME: A Novel Evaluation Metric for Video Understanding Intelligence

Developing video understanding intelligence is quite challenging because it requires holistic integration of images, scripts, and sounds based on natural language processing, temporal dependency, and reasoning. Recently, substantial attempts have been made on several video datasets with associated question answering (QA) on a large scale. However, existing evaluation metrics for video question answering (VideoQA) do not provide meaningful analysis. To make progress, we argue that a well-made framework, established on the way humans understand, is required to explain and evaluate the performance of understanding in detail. Then we propose a top-down evaluation system for VideoQA, based on the cognitive process of humans and story elements: Cognitive Modules for Evaluation (CogME). CogME is composed of three cognitive modules: targets, contents, and thinking. The interaction among the modules in the understanding procedure can be expressed in one sentence as follows: "I understand the CONTENT of the TARGET through a way of THINKING." Each module has sub-components derived from the story elements. We can specify the required aspects of understanding by annotating the sub-components to individual questions. CogME thus provides a framework for an elaborated specification of VideoQA datasets. To examine the suitability of a VideoQA dataset for validating video understanding intelligence, we evaluated the baseline model of the DramaQA dataset by applying CogME. The evaluation reveals that story elements are unevenly reflected in the existing dataset, and the model based on the dataset may cause biased predictions. Although this study has only been able to grasp a narrow range of stories, we expect that it offers the first step in considering the cognitive process of humans on the video understanding intelligence of humans and AI.

Towards a Federated Learning Framework for Heterogeneous Devices of Internet of Things

Federated Learning (FL) has received a significant amount of attention in the industry and research community due to its capability of keeping data on local devices. To aggregate the gradients of local models to train the global model, existing works require that the global model and the local models are the same. However, Internet of Things (IoT) devices are inherently diverse regarding computation speed and onboard memory. In this paper, we propose an FL framework targeting the heterogeneity of IoT devices. Specifically, local models are compressed from the global model, and the gradients of the compressed local models are used to update the global model. We conduct preliminary experiments to illustrate that our framework can facilitate the design of IoT-aware FL.