Yongkweon Jeon

Yongkweon Jeon, Chungman Lee, Ho-young Kim

Zero-shot quantization is a promising approach for developing lightweight deep neural networks when data is inaccessible owing to various reasons, including cost and issues related to privacy. By exploiting the learned parameters ($μ$ and $σ$) of batch normalization layers in an FP32-pre-trained model, zero-shot quantization schemes focus on generating synthetic data. Subsequently, they distill knowledge from the pre-trained model (teacher) to the quantized model (student) such that the quantized model can be optimized with the synthetic dataset. However, thus far, zero-shot quantization has primarily been discussed in the context of quantization-aware training methods, which require task-specific losses and long-term optimization as much as retraining. We thus introduce a post-training quantization scheme for zero-shot quantization that produces high-quality quantized networks within a few hours. Furthermore, we propose a framework called Genie~that generates data suited for quantization. With the data synthesized by Genie, we can produce robust quantized models without real datasets, which is comparable to few-shot quantization. We also propose a post-training quantization algorithm to enhance the performance of quantized models. By combining them, we can bridge the gap between zero-shot and few-shot quantization while significantly improving the quantization performance compared to that of existing approaches. In other words, we can obtain a unique state-of-the-art zero-shot quantization approach. The code is available at \url{https://github.com/SamsungLabs/Genie}.

Jinwoo Ahn, Ingyu Seong, Akhil Kedia et al.

Transformer-based large language models (LLMs) rely on key-value (KV) caching to avoid redundant computation during autoregressive inference. While this mechanism greatly improves efficiency, the cache size grows linearly with the input sequence length, quickly becoming a bottleneck for long-context tasks. Existing solutions mitigate this problem by evicting prompt KV that are deemed unimportant, guided by estimated importance scores. Notably, a recent line of work proposes to improve eviction quality by "glimpsing into the future", in which a draft generator produces a surrogate future response approximating the target model's true response, and this surrogate is subsequently used to estimate the importance of cached KV more accurately. However, these approaches rely on computationally expensive draft generation, which introduces substantial prefilling overhead and limits their practicality in real-world deployment. To address this challenge, we propose LookaheadKV, a lightweight eviction framework that leverages the strength of surrogate future response without requiring explicit draft generation. LookaheadKV augments transformer layers with parameter-efficient modules trained to predict true importance scores with high accuracy. Our design ensures negligible runtime overhead comparable to existing inexpensive heuristics, while achieving accuracy superior to more costly approximation methods. Extensive experiments on long-context understanding benchmarks, across a wide range of models, demonstrate that our method not only outperforms recent competitive baselines in various long-context understanding tasks, but also reduces the eviction cost by up to 14.5x, leading to significantly faster time-to-first-token. Our code is available at https://github.com/SamsungLabs/LookaheadKV.

Yeo Jeong Park, Hyemi Jang, Minseo Choi et al.

Omni-modal large language models have demonstrated remarkable potential in holistic multimodal understanding; however, the token explosion caused by high-resolution audio and video inputs remains a critical bottleneck for real-time applications and long-form reasoning. Existing omni-modal token compression methods typically prune tokens at the input embedding level, relying on audio-video similarity or temporal co-occurrence as proxies for semantic relevance. In practice, such assumptions are often unreliable. To address this limitation, we propose OmniDrop, a training-free, layer-wise token pruning framework that progressively prunes audiovisual tokens within the LLM decoder layers rather than at the input-level, allowing early layers to preserve sufficient omni-modal information fusion before aggressively removing tokens in deeper layers. We further utilize text queries as guidance for modality-agnostic and task-adaptive token pruning. We also introduce a temporal diversity score that encourages balanced token survival to preserve global temporal context. Experimental results across various audiovisual benchmarks demonstrate that OmniDrop outperforms all baselines by up to 3.58 points while reducing prefill latency by up to 40% and memory usage by up to 14.7%.

Junhan Kim, Yeo Jeong Park, Seungwoo Son et al.

The rapid growth of large language models (LLMs) has heightened the importance of post-training quantization (PTQ) for reducing memory and computation costs. Among PTQ methods, GPTQ has gained significant attention for its efficiency, enabling billion-scale LLMs to be quantized within a few GPU hours. However, GPTQ's assumption of layer-wise independence leads to severe accuracy drops in low-bit regimes. Recently, BoA improved upon GPTQ by incorporating inter-layer dependencies within attention modules, but its reliance on sequential quantization across all out-channels makes it substantially less efficient. In this paper, we propose TurboBoA, a new backpropagation-free PTQ algorithm that preserves the accuracy benefits of BoA while significantly accelerating the process. The proposed TurboBoA introduces three key innovations: (i) joint quantization of multiple out-channels with a closed-form error compensation rule, which reduces sequential bottlenecks and yields more than a three-fold speedup; (ii) a correction mechanism for errors propagated from preceding quantized layers; and (iii) adaptive grid computation with coordinate descent refinement to maintain alignment during iterative updates. Extensive experiments demonstrate that TurboBoA delivers substantial acceleration over BoA while consistently improving accuracy. When combined with outlier suppression techniques, it achieves state-of-the-art results in both weight-only and weight-activation quantization. The code will be available at https://github.com/SamsungLabs/TurboBoA.

Junhan Kim, Ho-young Kim, Eulrang Cho et al.

Post-training quantization (PTQ) is a promising solution for deploying large language models (LLMs) on resource-constrained devices. Early methods developed for small-scale networks, such as ResNet, rely on gradient-based optimization, which becomes impractical for hyper-scale LLMs with billions of parameters. While recently proposed backpropagation-free or transformation-based methods alleviate this issue, they ignore inter-layer interactions or use the naive nearest-rounding-based quantized weight assignment to save the heavy computational cost of weight optimization. In this paper, we introduce a novel backpropagation-free PTQ algorithm that optimizes quantized weights by considering inter-layer dependencies. The key innovation is the development of attention-aware Hessian matrices that capture inter-layer interactions within the attention module. Extensive experiments demonstrate that our approach not only outperforms existing weight quantization methods but also shows good synergy with conventional methods to suppress activation outliers, leading to state-of-the-art weight-activation quantization performance. The code will be available at https://github.com/SamsungLabs/BoA.

Junhan Kim, Gukryeol Lee, Seungwoo Son et al.

Group-wise quantization is an effective strategy for mitigating accuracy degradation in low-bit quantization of large language models (LLMs). Among existing methods, GPTQ has been widely adopted due to its efficiency; however, it neglects input statistics and inter-group correlations when determining group scales, leading to a mismatch with its goal of minimizing layer-wise reconstruction loss. In this work, we propose a two-stage optimization framework for group scales that explicitly minimizes the layer-wise reconstruction loss. In the first stage, performed prior to GPTQ, we initialize each group scale to minimize the group-wise reconstruction loss, thereby incorporating input statistics. In the second stage, we freeze the integer weights obtained via GPTQ and refine the group scales to minimize the layer-wise reconstruction loss. To this end, we employ the coordinate descent algorithm and derive a closed-form update rule, which enables efficient refinement without costly numerical optimization. Notably, our derivation incorporates the quantization errors from preceding layers to prevent error accumulation. Experimental results demonstrate that our method consistently enhances group-wise quantization, achieving higher accuracy with negligible overhead.

Junhan Kim, Chungman Lee, Eulrang Cho et al.

With the increasing complexity of generative AI models, post-training quantization (PTQ) has emerged as a promising solution for deploying hyper-scale models on edge devices such as mobile and TVs. Existing PTQ schemes, however, consume considerable time and resources, which could be a bottleneck in real situations where frequent model updates and multiple hyperparameter tunings are required. As a cost-effective alternative, learning-free PTQ schemes have been proposed. However, the performance is somewhat limited because they cannot consider the inter-layer dependency within the attention module, which is a significant feature of Transformers. In this paper, we thus propose a novel PTQ algorithm that balances accuracy and efficiency. The key idea of the proposed algorithm called aespa is to perform quantization layer-wise for efficiency while targeting attention-wise reconstruction to consider the cross-layer dependency. Through extensive experiments on various language models and complexity analysis, we demonstrate that aespa is accurate and efficient in quantizing Transformer models.

Dongsoo Lee, Se Jung Kwon, Byeongwook Kim et al.

While model compression is increasingly important because of large neural network size, compression-aware training is challenging as it needs sophisticated model modifications and longer training time.In this paper, we introduce regularization frequency (i.e., how often compression is performed during training) as a new regularization technique for a practical and efficient compression-aware training method. For various regularization techniques, such as weight decay and dropout, optimizing the regularization strength is crucial to improve generalization in Deep Neural Networks (DNNs). While model compression also demands the right amount of regularization, the regularization strength incurred by model compression has been controlled only by compression ratio. Throughout various experiments, we show that regularization frequency critically affects the regularization strength of model compression. Combining regularization frequency and compression ratio, the amount of weight updates by model compression per mini-batch can be optimized to achieve the best model accuracy. Modulating regularization frequency is implemented by occasional model compression while conventional compression-aware training is usually performed for every mini-batch.

Byeongwook Kim, Dongsoo Lee, Yeonju Ro et al.

Various post-training uniform quantization methods have usually been studied based on convex optimization. As a result, most previous ones rely on the quantization error minimization and/or quadratic approximations. Such approaches are computationally efficient and reasonable when a large number of quantization bits are employed. When the number of quantization bits is relatively low, however, non-convex optimization is unavoidable to improve model accuracy. In this paper, we propose a new post-training uniform quantization technique considering non-convexity. We empirically show that hyper-parameters for clipping and rounding of weights and activations can be explored by monitoring task loss. Then, an optimally searched set of hyper-parameters is frozen to proceed to the next layer such that an incremental non-convex optimization is enabled for post-training quantization. Throughout extensive experimental results using various models, our proposed technique presents higher model accuracy, especially for a low-bit quantization.

Insoo Chung, Byeongwook Kim, Yoonjung Choi et al.

The deployment of widely used Transformer architecture is challenging because of heavy computation load and memory overhead during inference, especially when the target device is limited in computational resources such as mobile or edge devices. Quantization is an effective technique to address such challenges. Our analysis shows that for a given number of quantization bits, each block of Transformer contributes to translation quality and inference computations in different manners. Moreover, even inside an embedding block, each word presents vastly different contributions. Correspondingly, we propose a mixed precision quantization strategy to represent Transformer weights by an extremely low number of bits (e.g., under 3 bits). For example, for each word in an embedding block, we assign different quantization bits based on statistical property. Our quantized Transformer model achieves 11.8$\times$ smaller model size than the baseline model, with less than -0.5 BLEU. We achieve 8.3$\times$ reduction in run-time memory footprints and 3.5$\times$ speed up (Galaxy N10+) such that our proposed compression strategy enables efficient implementation for on-device NMT.

Dongsoo Lee, Se Jung Kwon, Byeongwook Kim et al.

Quantization based on the binary codes is gaining attention because each quantized bit can be directly utilized for computations without dequantization using look-up tables. Previous attempts, however, only allow for integer numbers of quantization bits, which ends up restricting the search space for compression ratio and accuracy. In this paper, we propose an encryption algorithm/architecture to compress quantized weights so as to achieve fractional numbers of bits per weight. Decryption during inference is implemented by digital XOR-gate networks added into the neural network model while XOR gates are described by utilizing $\tanh(x)$ for backward propagation to enable gradient calculations. We perform experiments using MNIST, CIFAR-10, and ImageNet to show that inserting XOR gates learns quantization/encrypted bit decisions through training and obtains high accuracy even for fractional sub 1-bit weights. As a result, our proposed method yields smaller size and higher model accuracy compared to binary neural networks.

Yongkweon Jeon, Baeseong Park, Se Jung Kwon et al.

The number of parameters in deep neural networks (DNNs) is rapidly increasing to support complicated tasks and to improve model accuracy. Correspondingly, the amount of computations and required memory footprint increase as well. Quantization is an efficient method to address such concerns by compressing DNNs such that computations can be simplified while required storage footprint is significantly reduced. Unfortunately, commercial CPUs and GPUs do not fully support quantization because only fixed data transfers (such as 32 bits) are allowed. As a result, even if weights are quantized into a few bits, CPUs and GPUs cannot access multiple quantized weights without memory bandwidth waste. Success of quantization in practice, hence, relies on an efficient computation engine design, especially for matrix multiplication that is a basic computation engine in most DNNs. In this paper, we propose a novel matrix multiplication method, called BiQGEMM, dedicated to quantized DNNs. BiQGEMM can access multiple quantized weights simultaneously in one instruction. In addition, BiQGEMM pre-computes intermediate results that are highly redundant when quantization leads to limited available computation space. Since pre-computed values are stored in lookup tables and reused, BiQGEMM achieves lower amount of overall computations. Our extensive experimental results show that BiQGEMM presents higher performance than conventional schemes when DNNs are quantized.