Yongseok Choi

Sung Jun Cheon, Jaekyung Cho, Seongho Choi et al.

We introduce A.X K1, a 519B-parameter Mixture-of-Experts (MoE) language model trained from scratch. Our design leverages scaling laws to optimize training configurations and vocabulary size under fixed computational budgets. A.X K1 is pre-trained on a corpus of approximately 10T tokens, curated by a multi-stage data processing pipeline. Designed to bridge the gap between reasoning capability and inference efficiency, A.X K1 supports explicitly controllable reasoning to facilitate scalable deployment across diverse real-world scenarios. We propose a simple yet effective Think-Fusion training recipe, enabling user-controlled switching between thinking and non-thinking modes within a single unified model. Extensive evaluations demonstrate that A.X K1 achieves performance competitive with leading open-source models, while establishing a distinctive advantage in Korean-language benchmarks.

Wonjin Jung, Yongseok Choi

We address polymer property prediction with a multi-view design that exploits complementary representations. Our system integrates four families: (i) tabular RDKit/Morgan descriptors, (ii) graph neural networks, (iii) 3D-informed representations, and (iv) pretrained SMILES language models, and averages per-property predictions via a uniform ensemble. Models are trained with 10-fold splits and evaluated with SMILES test-time augmentation. The approach ranks 9th of 2241 teams in the Open Polymer Prediction Challenge at NeurIPS 2025. The submitted ensemble achieves a public MAE of 0.057 and a private MAE of 0.082.

Jiwoong Park, Chaeun Lee, Yongseok Choi et al.

Post-training quantization (PTQ) improves efficiency but struggles with limited calibration data, especially under privacy constraints. Data-free quantization (DFQ) mitigates this by generating synthetic images using generative models such as generative adversarial networks (GANs) and text-conditioned latent diffusion models (LDMs), while applying existing PTQ algorithms. However, the relationship between generated synthetic images and the generalizability of the quantized model during PTQ remains underexplored. Without investigating this relationship, synthetic images generated by previous prompt engineering methods based on single-class prompts suffer from issues such as polysemy, leading to performance degradation. We propose \textbf{mixup-class prompt}, a mixup-based text prompting strategy that fuses multiple class labels at the text prompt level to generate diverse, robust synthetic data. This approach enhances generalization, and improves optimization stability in PTQ. We provide quantitative insights through gradient norm and generalization error analysis. Experiments on convolutional neural networks (CNNs) and vision transformers (ViTs) show that our method consistently outperforms state-of-the-art DFQ methods like GenQ. Furthermore, it pushes the performance boundary in extremely low-bit scenarios, achieving new state-of-the-art accuracy in challenging 2-bit weight, 4-bit activation (W2A4) quantization.

Yunho Jeon, Yongseok Choi, Jaesun Park et al.

Training a deep neural network with a small amount of data is a challenging problem as it is vulnerable to overfitting. However, one of the practical difficulties that we often face is to collect many samples. Transfer learning is a cost-effective solution to this problem. By using the source model trained with a large-scale dataset, the target model can alleviate the overfitting originated from the lack of training data. Resorting to the ability of generalization of the source model, several methods proposed to use the source knowledge during the whole training procedure. However, this is likely to restrict the potential of the target model and some transferred knowledge from the source can interfere with the training procedure. For improving the generalization performance of the target model with a few training samples, we proposed a regularization method called sample-based regularization (SBR), which does not rely on the source's knowledge during training. With SBR, we suggested a new training framework for transfer learning. Experimental results showed that our framework outperformed existing methods in various configurations.

Jin-Hwa Kim, Junyoung Park, Yongseok Choi

Gradient-based meta-learning approaches have been successful in few-shot learning, transfer learning, and a wide range of other domains. Despite its efficacy and simplicity, the burden of calculating the Hessian matrix with large memory footprints is the critical challenge in large-scale applications. To tackle this issue, we propose a simple yet straightforward method to reduce the cost by reusing the same gradient in a window of inner steps. We describe the dynamics of the multi-step estimation in the Lagrangian formalism and discuss how to reduce evaluating second-order derivatives estimating the dynamics. To validate our method, we experiment on meta-transfer learning and few-shot learning tasks for multiple settings. The experiment on meta-transfer emphasizes the applicability of training meta-networks, where other approximations are limited. For few-shot learning, we evaluate time and memory complexities compared with popular baselines. We show that our method significantly reduces training time and memory usage, maintaining competitive accuracies, or even outperforming in some cases.

Yongseok Choi, Junyoung Park, Subin Yi et al.

Although few-shot learning research has advanced rapidly with the help of meta-learning, its practical usefulness is still limited because most of them assumed that all meta-training and meta-testing examples came from a single domain. We propose a simple but effective way for few-shot classification in which a task distribution spans multiple domains including ones never seen during meta-training. The key idea is to build a pool of models to cover this wide task distribution and learn to select the best one for a particular task through cross-domain meta-learning. All models in the pool share a base network while each model has a separate modulator to refine the base network in its own way. This framework allows the pool to have representational diversity without losing beneficial domain-invariant features. We verify the effectiveness of the proposed algorithm through experiments on various datasets across diverse domains.

Junyoung Park, Subin Yi, Yongseok Choi et al.

Metric-based few-shot learning methods try to overcome the difficulty due to the lack of training examples by learning embedding to make comparison easy. We propose a novel algorithm to generate class representatives for few-shot classification tasks. As a probabilistic model for learned features of inputs, we consider a mixture of von Mises-Fisher distributions which is known to be more expressive than Gaussian in a high dimensional space. Then, from a discriminative classifier perspective, we get a better class representative considering inter-class correlation which has not been addressed by conventional few-shot learning algorithms. We apply our method to \emph{mini}ImageNet and \emph{tiered}ImageNet datasets, and show that the proposed approach outperforms other comparable methods in few-shot classification tasks.

Jaehong Kim, Sungeun Hong, Yongseok Choi et al.

We propose doubly nested network(DNNet) where all neurons represent their own sub-models that solve the same task. Every sub-model is nested both layer-wise and channel-wise. While nesting sub-models layer-wise is straight-forward with deep-supervision as proposed in \cite{xie2015holistically}, channel-wise nesting has not been explored in the literature to our best knowledge. Channel-wise nesting is non-trivial as neurons between consecutive layers are all connected to each other. In this work, we introduce a technique to solve this problem by sorting channels topologically and connecting neurons accordingly. For the purpose, channel-causal convolutions are used. Slicing doubly nested network gives a working sub-network. The most notable application of our proposed network structure with slicing operation is resource-efficient inference. At test time, computing resources such as time and memory available for running the prediction algorithm can significantly vary across devices and applications. Given a budget constraint, we can slice the network accordingly and use a sub-model for inference within budget, requiring no additional computation such as training or fine-tuning after deployment. We demonstrate the effectiveness of our approach in several practical scenarios of utilizing available resource efficiently.

Jaehong Kim, Sangyeul Lee, Sungwan Kim et al.

Fully automating machine learning pipelines is one of the key challenges of current artificial intelligence research, since practical machine learning often requires costly and time-consuming human-powered processes such as model design, algorithm development, and hyperparameter tuning. In this paper, we verify that automated architecture search synergizes with the effect of gradient-based meta learning. We adopt the progressive neural architecture search \cite{liu:pnas_google:DBLP:journals/corr/abs-1712-00559} to find optimal architectures for meta-learners. The gradient based meta-learner whose architecture was automatically found achieved state-of-the-art results on the 5-shot 5-way Mini-ImageNet classification problem with $74.65\%$ accuracy, which is $11.54\%$ improvement over the result obtained by the first gradient-based meta-learner called MAML \cite{finn:maml:DBLP:conf/icml/FinnAL17}. To our best knowledge, this work is the first successful neural architecture search implementation in the context of meta learning.