Peiqi Wang

Tencent Hunyuan Team, Ao Liu, Botong Zhou et al. · tencent-ai

As Large Language Models (LLMs) rapidly advance, we introduce Hunyuan-TurboS, a novel large hybrid Transformer-Mamba Mixture of Experts (MoE) model. It synergistically combines Mamba's long-sequence processing efficiency with Transformer's superior contextual understanding. Hunyuan-TurboS features an adaptive long-short chain-of-thought (CoT) mechanism, dynamically switching between rapid responses for simple queries and deep "thinking" modes for complex problems, optimizing computational resources. Architecturally, this 56B activated (560B total) parameter model employs 128 layers (Mamba2, Attention, FFN) with an innovative AMF/MF block pattern. Faster Mamba2 ensures linear complexity, Grouped-Query Attention minimizes KV cache, and FFNs use an MoE structure. Pre-trained on 16T high-quality tokens, it supports a 256K context length and is the first industry-deployed large-scale Mamba model. Our comprehensive post-training strategy enhances capabilities via Supervised Fine-Tuning (3M instructions), a novel Adaptive Long-short CoT Fusion method, Multi-round Deliberation Learning for iterative improvement, and a two-stage Large-scale Reinforcement Learning process targeting STEM and general instruction-following. Evaluations show strong performance: overall top 7 rank on LMSYS Chatbot Arena with a score of 1356, outperforming leading models like Gemini-2.0-Flash-001 (1352) and o4-mini-2025-04-16 (1345). TurboS also achieves an average of 77.9% across 23 automated benchmarks. Hunyuan-TurboS balances high performance and efficiency, offering substantial capabilities at lower inference costs than many reasoning models, establishing a new paradigm for efficient large-scale pre-trained models.

Peiqi Wang, William M. Wells, Seth Berkowitz et al.

Image-text multimodal representation learning aligns data across modalities and enables important medical applications, e.g., image classification, visual grounding, and cross-modal retrieval. In this work, we establish a connection between multimodal representation learning and multiple instance learning. Based on this connection, we propose a generic framework for constructing permutation-invariant score functions with many existing multimodal representation learning approaches as special cases. Furthermore, we use the framework to derive a novel contrastive learning approach and demonstrate that our method achieves state-of-the-art results in several downstream tasks.

Peiqi Wang, Yingcheng Liu, Ching-Yun Ko et al.

Self-supervised representation learning on image-text data facilitates crucial medical applications, such as image classification, visual grounding, and cross-modal retrieval. One common approach involves contrasting semantically similar (positive) and dissimilar (negative) pairs of data points. Drawing negative samples uniformly from the training data set introduces false negatives, i.e., samples that are treated as dissimilar but belong to the same class. In healthcare data, the underlying class distribution is nonuniform, implying that false negatives occur at a highly variable rate. To improve the quality of learned representations, we develop a novel approach that corrects for false negatives. Our method can be viewed as a variant of debiased contrastive learning that uses estimated sample-specific class probabilities. We provide theoretical analysis of the objective function and demonstrate the proposed approach on both image and paired image-text data sets. Our experiments illustrate empirical advantages of sample-specific debiasing.

Yingcheng Liu, Peiqi Wang, Sebastian Diaz et al.

Analyzing fetal body motion and shape is paramount in prenatal diagnostics and monitoring. Existing methods for fetal MRI analysis mainly rely on anatomical keypoints or volumetric body segmentations. Keypoints simplify body structure to facilitate motion analysis, but may ignore important details of full-body shape. Body segmentations capture complete shape information but complicate temporal analysis due to large non-local fetal movements. To address these limitations, we construct a 3D articulated statistical fetal body model based on the Skinned Multi-Person Linear Model (SMPL). Our algorithm iteratively estimates body pose in the image space and body shape in the canonical pose space. This approach improves robustness to MRI motion artifacts and intensity distortions, and reduces the impact of incomplete surface observations due to challenging fetal poses. We train our model on segmentations and keypoints derived from $19,816$ MRI volumes across $53$ subjects. Our model captures body shape and motion across time series and provides intuitive visualization. Furthermore, it enables automated anthropometric measurements traditionally difficult to obtain from segmentations and keypoints. When tested on unseen fetal body shapes, our method yields a surface alignment error of $3.2$ mm for $3$ mm MRI voxel size. To our knowledge, this represents the first 3D articulated statistical fetal body model, paving the way for enhanced fetal motion and shape analysis in prenatal diagnostics. The code is available at https://github.com/MedicalVisionGroup/fetal-smpl .

Peiqi Wang, Yikang Shen, Zhen Guo et al.

We aim to select data subsets for the fine-tuning of large language models to more effectively follow instructions. Prior work has emphasized the importance of diversity in dataset curation but relied on heuristics such as the number of tasks. In this paper, we use determinantal point processes to capture the diversity and quality of instruction tuning datasets for subset selection. We propose to measure dataset diversity with log determinant distance that is the distance between the dataset of interest and a maximally diverse reference dataset. Our experiments demonstrate that the proposed diversity measure in the normalized weight gradient space is correlated with downstream instruction-following performance. Consequently, it can be used to inform when data selection is the most helpful and to analyze dataset curation strategies. We demonstrate the utility of our approach on various instruction tuning datasets.

Peiqi Wang, ShengYun Peng, Xuewen Zhang et al.

This work investigates the optimal allocation of inference compute across three key scaling factors in video vision language models: language model size, frame count, and the number of visual tokens per frame. While prior works typically focuses on optimizing model efficiency or improving performance without considering resource constraints, we instead identify optimal model configuration under fixed inference compute budgets. We conduct large-scale training sweeps and careful parametric modeling of task performance to identify the inference compute-optimal frontier. Our experiments reveal how task performance depends on scaling factors and finetuning data size, as well as how changes in data size shift the compute-optimal frontier. These findings translate to practical tips for selecting these scaling factors.

Zhen Guo, Peiqi Wang, Yanwei Wang et al.

Large Language Models (LLMs) have made remarkable advancements in the field of natural language processing. However, their increasing size poses challenges in terms of computational cost. On the other hand, Small Language Models (SLMs) are known for their efficiency, but they often struggle with limited capacity and training data, especially in specific domains. In this paper, we introduce a novel method aimed at improving SLMs in the medical domain using LLM-based generative data augmentation. The objective of our approach is to develop more efficient and capable models that are specifically tailored for specialized applications. Through experiments conducted on the PubMedQA dataset, we demonstrate the effectiveness of LLMs in refining and diversifying existing question-answer pairs. This refinement process leads to improved performance in a significantly smaller model after fine-tuning. Notably, our best SLM, with under 1.6 billion parameters, outperforms the few-shot GPT-4 on the PubMedQA dataset. Our code and generated data are publicly available to facilitate further explorations.

Peiqi Wang, Ruizhi Liao, Daniel Moyer et al.

Adoption of machine learning models in healthcare requires end users' trust in the system. Models that provide additional supportive evidence for their predictions promise to facilitate adoption. We define consistent evidence to be both compatible and sufficient with respect to model predictions. We propose measures of model inconsistency and regularizers that promote more consistent evidence. We demonstrate our ideas in the context of edema severity grading from chest radiographs. We demonstrate empirically that consistent models provide competitive performance while supporting interpretation.

Yu Ji, Youyang Zhang, Xinfeng Xie et al.

Neural Network (NN) accelerators with emerging ReRAM (resistive random access memory) technologies have been investigated as one of the promising solutions to address the \textit{memory wall} challenge, due to the unique capability of \textit{processing-in-memory} within ReRAM-crossbar-based processing elements (PEs). However, the high efficiency and high density advantages of ReRAM have not been fully utilized due to the huge communication demands among PEs and the overhead of peripheral circuits. In this paper, we propose a full system stack solution, composed of a reconfigurable architecture design, Field Programmable Synapse Array (FPSA) and its software system including neural synthesizer, temporal-to-spatial mapper, and placement & routing. We highly leverage the software system to make the hardware design compact and efficient. To satisfy the high-performance communication demand, we optimize it with a reconfigurable routing architecture and the placement & routing tool. To improve the computational density, we greatly simplify the PE circuit with the spiking schema and then adopt neural synthesizer to enable the high density computation-resources to support different kinds of NN operations. In addition, we provide spiking memory blocks (SMBs) and configurable logic blocks (CLBs) in hardware and leverage the temporal-to-spatial mapper to utilize them to balance the storage and computation requirements of NN. Owing to the end-to-end software system, we can efficiently deploy existing deep neural networks to FPSA. Evaluations show that, compared to one of state-of-the-art ReRAM-based NN accelerators, PRIME, the computational density of FPSA improves by 31x; for representative NNs, its inference performance can achieve up to 1000x speedup.

Peiqi Wang, Dongsheng Wang, Yu Ji et al.

The intensive computation and memory requirements of generative adversarial neural networks (GANs) hinder its real-world deployment on edge devices such as smartphones. Despite the success in model reduction of CNNs, neural network quantization methods have not yet been studied on GANs, which are mainly faced with the issues of both the effectiveness of quantization algorithms and the instability of training GAN models. In this paper, we start with an extensive study on applying existing successful methods to quantize GANs. Our observation reveals that none of them generates samples with reasonable quality because of the underrepresentation of quantized values in model weights, and the generator and discriminator networks show different sensitivities upon quantization methods. Motivated by these observations, we develop a novel quantization method for GANs based on EM algorithms, named as QGAN. We also propose a multi-precision algorithm to help find the optimal number of bits of quantized GAN models in conjunction with corresponding result qualities. Experiments on CIFAR-10 and CelebA show that QGAN can quantize GANs to even 1-bit or 2-bit representations with results of quality comparable to original models.

Yu Ji, Zixin Liu, Xing Hu et al.

Neural network (NN) trojaning attack is an emerging and important attack model that can broadly damage the system deployed with NN models. Existing studies have explored the outsourced training attack scenario and transfer learning attack scenario in some small datasets for specific domains, with limited numbers of fixed target classes. In this paper, we propose a more powerful trojaning attack method for both outsourced training attack and transfer learning attack, which outperforms existing studies in the capability, generality, and stealthiness. First, The attack is programmable that the malicious misclassification target is not fixed and can be generated on demand even after the victim's deployment. Second, our trojan attack is not limited in a small domain; one trojaned model on a large-scale dataset can affect applications of different domains that reuse its general features. Thirdly, our trojan design is hard to be detected or eliminated even if the victims fine-tune the whole model.