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.

No Token Left Behind: Reliable KV Cache Compression via Importance-Aware Mixed Precision Quantization

Key-Value (KV) Caching has become an essential technique for accelerating the inference speed and throughput of generative Large Language Models~(LLMs). However, the memory footprint of the KV cache poses a critical bottleneck in LLM deployment as the cache size grows with batch size and sequence length, often surpassing even the size of the model itself. Although recent methods were proposed to select and evict unimportant KV pairs from the cache to reduce memory consumption, the potential ramifications of eviction on the generative process are yet to be thoroughly examined. In this paper, we examine the detrimental impact of cache eviction and observe that unforeseen risks arise as the information contained in the KV pairs is exhaustively discarded, resulting in safety breaches, hallucinations, and context loss. Surprisingly, we find that preserving even a small amount of information contained in the evicted KV pairs via reduced precision quantization substantially recovers the incurred degradation. On the other hand, we observe that the important KV pairs must be kept at a relatively higher precision to safeguard the generation quality. Motivated by these observations, we propose \textit{Mixed-precision KV cache}~(MiKV), a reliable cache compression method that simultaneously preserves the context details by retaining the evicted KV pairs in low-precision and ensure generation quality by keeping the important KV pairs in high-precision. Experiments on diverse benchmarks and LLM backbones show that our proposed method offers a state-of-the-art trade-off between compression ratio and performance, compared to other baselines.

DropBP: Accelerating Fine-Tuning of Large Language Models by Dropping Backward Propagation

Training deep neural networks typically involves substantial computational costs during both forward and backward propagation. The conventional layer dropping techniques drop certain layers during training for reducing the computations burden. However, dropping layers during forward propagation adversely affects the training process by degrading accuracy. In this paper, we propose Dropping Backward Propagation (DropBP), a novel approach designed to reduce computational costs while maintaining accuracy. DropBP randomly drops layers during the backward propagation, which does not deviate forward propagation. Moreover, DropBP calculates the sensitivity of each layer to assign appropriate drop rate, thereby stabilizing the training process. DropBP is designed to enhance the efficiency of the training process with backpropagation, thereby enabling the acceleration of both full fine-tuning and parameter-efficient fine-tuning using backpropagation. Specifically, utilizing DropBP in QLoRA reduces training time by 44%, increases the convergence speed to the identical loss level by 1.5$\times$, and enables training with a 6.2$\times$ larger sequence length on a single NVIDIA-A100 80GiB GPU in LLaMA2-70B. The code is available at https://github.com/WooSunghyeon/dropbp.

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.

DFX: A Low-latency Multi-FPGA Appliance for Accelerating Transformer-based Text Generation

Transformer is a deep learning language model widely used for natural language processing (NLP) services in datacenters. Among transformer models, Generative Pre-trained Transformer (GPT) has achieved remarkable performance in text generation, or natural language generation (NLG), which needs the processing of a large input context in the summarization stage, followed by the generation stage that produces a single word at a time. The conventional platforms such as GPU are specialized for the parallel processing of large inputs in the summarization stage, but their performance significantly degrades in the generation stage due to its sequential characteristic. Therefore, an efficient hardware platform is required to address the high latency caused by the sequential characteristic of text generation. In this paper, we present DFX, a multi-FPGA acceleration appliance that executes GPT-2 model inference end-to-end with low latency and high throughput in both summarization and generation stages. DFX uses model parallelism and optimized dataflow that is model-and-hardware-aware for fast simultaneous workload execution among devices. Its compute cores operate on custom instructions and provide GPT-2 operations end-to-end. We implement the proposed hardware architecture on four Xilinx Alveo U280 FPGAs and utilize all of the channels of the high bandwidth memory (HBM) and the maximum number of compute resources for high hardware efficiency. DFX achieves 5.58x speedup and 3.99x energy efficiency over four NVIDIA V100 GPUs on the modern GPT-2 model. DFX is also 8.21x more cost-effective than the GPU appliance, suggesting that it is a promising solution for text generation workloads in cloud datacenters.

nuQmm: Quantized MatMul for Efficient Inference of Large-Scale Generative Language Models

The recent advance of self-supervised learning associated with the Transformer architecture enables natural language processing (NLP) to exhibit extremely low perplexity. Such powerful models demand ever-increasing model size, and thus, large amounts of computations and memory footprints. In this paper, we propose an efficient inference framework for large-scale generative language models. As the key to reducing model size, we quantize weights by a non-uniform quantization method. Then, quantized matrix multiplications are accelerated by our proposed kernel, called nuQmm, which allows a wide trade-off between compression ratio and accuracy. Our proposed nuQmm reduces the latency of not only each GPU but also the entire inference of large LMs because a high compression ratio (by low-bit quantization) mitigates the minimum required number of GPUs. We demonstrate that nuQmm can accelerate the inference speed of the GPT-3 (175B) model by about 14.4 times and save energy consumption by 93%.