Author name cluster

Mingxuan Wang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

16 papers

2 author rows

TMLR Journal 2026 Journal Article

ACDiT: Interpolating Autoregressive Conditional Modeling and Diffusion Transformer

Jinyi Hu
Shengding Hu
Yuxuan Song
Yufei Huang
Mingxuan Wang
Hao Zhou
Zhiyuan Liu
Wei-Ying Ma

Autoregressive and diffusion models have achieved remarkable progress in language models and visual generation, respectively. We present ACDiT, a novel Autoregressive blockwise Conditional Diffusion Transformer, that innovatively combines autoregressive and diffusion paradigms for continuous visual information. By introducing a block-wise autoregressive unit, ACDiT offers a flexible interpolation between token-wise autoregression and full-sequence diffusion, bypassing the limitations of discrete tokenization. The generation of each block is formulated as a conditional diffusion process, conditioned on prior blocks. ACDiT is easy to implement, as simple as applying a specially designed Skip-Causal Attention Mask on the standard diffusion transformer during training. During inference, the process iterates between diffusion denoising and autoregressive decoding that can make full use of KV-Cache. We validate the effectiveness of ACDiT on image, video, and text generation and show that ACDiT performs best among all autoregressive baselines under similar model scales on visual generation tasks. We also demonstrate that, benefiting from autoregressive modeling, pretrained ACDiT can be transferred in visual understanding tasks despite being trained with the generative objective. The analysis of the trade-off between autoregressive and diffusion demonstrates the potential of ACDiT to be used in long-horizon visual generation tasks. We hope that ACDiT offers a novel perspective on visual autoregressive generation and sheds light on new avenues for unified models.

PDF Details

NeurIPS Conference 2025 Conference Paper

DAPO: An Open-Source LLM Reinforcement Learning System at Scale

Qiying Yu
Zheng Zhang
Ruofei Zhu
Yufeng Yuan
Xiaochen Zuo
Yu Yue
Weinan Dai
Tiantian Fan

Inference scaling empowers LLMs with unprecedented reasoning ability, with reinforcement learning as the core technique to elicit complex reasoning. However, key technical details of state-of-the-art reasoning LLMs are concealed (such as in OpenAI o1 blog and DeepSeek R1 technical report), thus the community still struggles to reproduce their RL training results. We propose the D ecoupled Clip and D ynamic s A mpling P olicy O ptimization ( DAPO ) algorithm, and fully open-source a state-of-the-art large-scale RL system that achieves 50 points on AIME 2024 using Qwen2. 5-32B base model. Unlike previous works that withhold training details, we introduce four key techniques of our algorithm that make large-scale LLM RL a success. In addition, we open-source our training code, which is built on the verl framework, along with a carefully curated and processed dataset. These components of our open-source system enhance reproducibility and support future research in large-scale LLM RL.

PDF Details

AAAI Conference 2025 Conference Paper

Empowering Self-Learning of LLMs: Inner Knowledge Explicitation as a Catalyst

Shijue Huang
Wanjun Zhong
Deng Cai
Fanqi Wan
Chengyi Wang
Mingxuan Wang
Mu Qiao
Ruifeng Xu

Self-learning of Large Language Models (LLMs) facilitates their advancement towards super-intelligence by training with self-synthesized experiences. However, a critical challenge is the amplification of hallucinations in generated data during iterative self-learning, underscoring the need for reliable data selection. To address this, we investigate the mechanism of Inner Knowledge Explicitation, which involves explicitly extracting the inner knowledge from memory of LLMs, to concurrently improves reasoning, and enables reliable self-learning data selection. This paper introduces a Self Knowledge Explicitation Learning (SKE-Learn) framework, which equips the LLMs with meta-skills to explicitly extract, verify and utilize inner knowledge for reasoning. By leveraging these meta-skills, SKE-Learn establishes a self-learning approach that ensures reliable selection of self-synthetic data. This approach enhances performance through iterative self-learning while mitigating the problem of hallucinations. Empirical results from six benchmarks demonstrate that Inner Knowledge Explicitation improves reasoning by serving as a more effective prompting method. Additionally, SKE-Learn, based on the verifiability of explicit knowledge, shows consistent performance improvements over multiple self-training iterations, with an average performance increase from 52.79% to 56.54% across all benchmarks. Furthermore, Inner Knowledge Explicitation provides explanation and intervention space during LLM's generation process.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

Enigmata: Scaling Logical Reasoning in Large Language Models with Synthetic Verifiable Puzzles

Jiangjie Chen
Qianyu He
Siyu Yuan
Aili Chen
Zhicheng Cai
Weinan Dai
Hongli Yu
Jiaze Chen

Large Language Models (LLMs), such as OpenAI’s o1 and DeepSeek’s R1, excel at advanced reasoning tasks like math and coding via Reinforcement Learning with Verifiable Rewards (RLVR), but still struggle with puzzles solvable by humans without domain knowledge. We introduce ENIGMATA, the first comprehensive suite tailored for improving LLMs with puzzle reasoning skills. It includes 36 tasks across 7 categories, each with: 1) a generator that produces unlimited examples with controllable difficulty, and 2) a rule-based verifier for automatic evaluation. This generator-verifier design supports scalable, multi-task RL training, fine-grained analysis, and seamless RLVR integration. We further propose ENIGMATA-Eval, a rigorous benchmark, and develop optimized multi-task RLVR strategies. Our trained model, Qwen2. 5-32B-ENIGMATA, consistently surpasses o3-mini-high and o1 on the puzzle reasoning benchmarks like ENIGMATA-Eval, ARC-AGI (32. 8%), and ARC-AGI 2 (0. 6%). It also generalizes well to out-of-domain puzzle benchmarks and mathematical reasoning, with little multi-tasking trade-off. When trained on larger models like Seed1. 5-Thinking (20B activated parameters and 200B total parameters), puzzle data from ENIGMATA further boosts SoTA performance on advanced math and STEM reasoning tasks such as AIME (2024-2025), BeyondAIME and GPQA (Diamond), showing nice generalization benefits of ENIGMATA. This work offers a unified, controllable framework for advancing logical reasoning in LLMs. Project page: https: //seed-enigmata. github. io.

PDF Details

AIIM Journal 2025 Journal Article

SCLResNet and DSAF: A self-supervised contrastive learning and deep self-attention fusion-based multimodal network for predicting central lymph node metastasis in papillary thyroid carcinoma

Shidi Miao
Yuyang Jiang
Wenjuan Huang
Yuxin Jiang
Mengzhuo Sun
Mingxuan Wang
Hongzhuo Qi
Ao Li

Accurate prediction of central lymph node metastasis (CLNM) in papillary thyroid carcinoma (PTC) is crucial to avoid unnecessary invasive procedures, yet existing models often fall short. We constructed the SCLResNet101 model based on a contrastive learning framework to extract network features of tumor ultrasound (US). SeResnet101 was used to extract network features of peri-vascular adipose tissue (PVAT) from the computed tomography (CT) of C6 (the arterial and venous layers beneath the thyroid). Univariate and multivariate analyses were performed using binary logistic regression to select clinical features. Finally, we constructed a Deep Self-Attention Fusion (DSAF) network to integrate features from these three modalities for CLNM prediction. Univariate and multivariate analyses revealed that Gender, Age, Size of US, and Extrathyroidal Extension (ETE) were independent risk factors for CLNM. In the internal test cohort (I-T), the area under the curve (AUC) of model was 0. 863 (95 % CI: 0. 779–0. 932). In the external test cohort (E-T), the AUC was 0. 839 (95 % CI: 0. 755–0. 905). Compared to all radiologists, the model significantly reduced both false-positive and false-negative rates in both the I-T and E-T. This study incorporates PVAT, which significantly enhances the performance of the multimodal deep learning model and may assist surgeons in making more informed and precise surgical decisions in the treatment of PTC.

Details DOI

NeurIPS Conference 2025 Conference Paper

ShortListing Model: A Streamlined Simplex Diffusion for Discrete Variable Generation

Yuxuan Song
Zhe Zhang
Yu Pei
Jingjing Gong
Qiying Yu
Zheng Zhang
Mingxuan Wang
Hao Zhou

Generative modeling of discrete variables is challenging yet crucial for applications in natural language processing and biological sequence design. We introduce the Shortlisting Model (SLM), a novel simplex-based diffusion model inspired by progressive candidate pruning. SLM operates on simplex centroids, reducing generation complexity and enhancing scalability. Additionally, SLM incorporates a flexible implementation of classifier-free guidance, enhancing unconditional generation performance. Extensive experiments on DNA promoter and enhancer design, protein design, character-level and large-vocabulary language modeling demonstrate the competitive performance and strong potential of SLM. Our code can be found at https: //github. com/GenSI-THUAIR/SLM.

PDF Details

ICLR Conference 2024 Conference Paper

PolyVoice: Language Models for Speech to Speech Translation

Qianqian Dong
Zhiying Huang
Qi Tian 0001
Chen Xu 0008
Tom Ko
Yunlong Zhao 0004
Siyuan Feng
Tang Li 0001

With the huge success of GPT models in natural language processing, there is a growing interest in applying language modeling approaches to speech tasks. Currently, the dominant architecture in speech-to-speech translation (S2ST) remains the encoder-decoder paradigm, creating a need to investigate the impact of language modeling approaches in this area. In this study, we introduce PolyVoice, a language model-based framework designed for S2ST systems. Our framework comprises three decoder-only language models: a translation language model, a duration language model, and a speech synthesis language model. These language models employ different types of prompts to extract learned information effectively. By utilizing unsupervised semantic units, our framework can transfer semantic information across these models, making it applicable even to unwritten languages. We evaluate our system on Chinese $\rightarrow$ English and English $\rightarrow$ Spanish language pairs. Experimental results demonstrate that \method outperforms the state-of-the-art encoder-decoder model, producing voice-cloned speech with high translation and audio quality. Speech samples are available at https://polyvoice.github.io.

Details

EAAI Journal 2024 Journal Article

Residual adaptive sparse hybrid attention transformer for image super resolution

Hai Huan
Mingxuan Wang

Image super-resolution is a visual task that reconstructs low-resolution images into high-resolution ones. Currently, many researchers favor applying Transformer-based methods to image super-resolution tasks, which have yielded promising results. However, due to the need to capture long-range dependencies across the entire image, existing research on Vision Transformers (ViT) for super-resolution reconstruction incurs high computational costs, thereby increasing system overhead. Additionally, some researchers have proposed methods based on manually sparse attention mechanisms. However, these approaches, which acquire receptive fields in a manner similar to traditional convolutions, do not fully exploit the advantages of Transformers in extracting global information, resulting in suboptimal reconstruction performance. To leverage the Transformer’s ability to capture long-range dependencies, this paper introduces a novel network called RASHAT. In RASHAT, we propose an Adaptive Sparse Hybrid Attention Block (ASHAB). This module introduces a Bi-level Routing Attention(BRA), incorporating both Channel Attention(CA) and Switch Window Multi-head Self-attention((S)W-MSA). These components are designed to capture long-range dependencies, global context, and local dependencies within the image. Additionally, the model employs an Overlapping Cross-Attention Block(OCAB) to enhance information interaction between neighboring pixels. During model training, we introduce a novel composite loss function that combines frequency domain loss with pixel loss, further improving model performance. Extensive experiments demonstrate that benefiting from the sparse attention provided by the Bi-Level Routing Attention (BRA), RASHAT achieves similar performance to the current state-of-the-art results (20. 8M) with significantly fewer parameters (11. 6M). These results hold across multiple commonly used datasets.

Details DOI

IJCAI Conference 2023 Conference Paper

Recent Advances in Direct Speech-to-text Translation

Chen Xu
Rong Ye
Qianqian Dong
Chengqi Zhao
Tom Ko
Mingxuan Wang
Tong Xiao
Jingbo Zhu

Recently, speech-to-text translation has attracted more and more attention and many studies have emerged rapidly. In this paper, we present a comprehensive survey on direct speech translation aiming to summarize the current state-of-the-art techniques. First, we categorize the existing research work into three directions based on the main challenges --- modeling burden, data scarcity, and application issues. To tackle the problem of modeling burden, two main structures have been proposed, encoder-decoder framework (Transformer and the variants) and multitask frameworks. For the challenge of data scarcity, recent work resorts to many sophisticated techniques, such as data augmentation, pre-training, knowledge distillation, and multilingual modeling. We analyze and summarize the application issues, which include real-time, segmentation, named entity, gender bias, and code-switching. Finally, we discuss some promising directions for future work.

PDF Details DOI

ICLR Conference 2022 Conference Paper

switch-GLAT: Multilingual Parallel Machine Translation Via Code-Switch Decoder

Zhenqiao Song
Hao Zhou 0012
Lihua Qian
Jingjing Xu 0001
Shanbo Cheng
Mingxuan Wang
Lei Li 0005

Multilingual machine translation aims to develop a single model for multiple language directions. However, existing multilingual models based on Transformer are limited in terms of both translation performance and inference speed. In this paper, we propose switch-GLAT, a non-autoregressive multilingual machine translation model with a code-switch decoder. It can generate contextual code-switched translations for a given source sentence, and perform code-switch back-translation, greatly boosting multilingual translation performance. In addition, its inference is highly efficient thanks to its parallel decoder. Experiments show that our proposed switch-GLAT outperform the multilingual Transformer with as much as 0.74 BLEU improvement and 6.2x faster decoding speed in inference.

Details

AAAI Conference 2021 Conference Paper

Consecutive Decoding for Speech-to-text Translation

Qianqian Dong
Mingxuan Wang
Hao Zhou
Shuang Xu
Bo Xu
Lei Li

Speech-to-text translation (ST), which directly translates the source language speech to the target language text, has attracted intensive attention recently. However, the combination of speech recognition and machine translation in a single model poses a heavy burden on the direct cross-modal crosslingual mapping. To reduce the learning difficulty, we propose COnSecutive Transcription and Translation (COSTT), an integral approach for speech-to-text translation. The key idea is to generate source transcript and target translation text with a single decoder. It benefits the model training so that additional large parallel text corpus can be fully exploited to enhance the speech translation training. Our method is verified on three mainstream datasets, including Augmented LibriSpeech English-French dataset, TED English-German dataset, and TED English-Chinese dataset. Experiments show that our proposed COSTT outperforms the previous state-ofthe-art methods. The code is available at https: //github. com/ dqqcasia/st.

PDF Details

AAAI Conference 2021 Conference Paper

Finding Sparse Structures for Domain Specific Neural Machine Translation

Jianze Liang
Chengqi Zhao
Mingxuan Wang
Xipeng Qiu
Lei Li

Neural machine translation often adopts the fine-tuning approach to adapt to specific domains. However, nonrestricted fine-tuning can easily degrade on the general domain and over-fit to the target domain. To mitigate the issue, we propose PRUNE-TUNE, a novel domain adaptation method via gradual pruning. It learns tiny domain-specific sub-networks during fine-tuning on new domains. PRUNE-TUNE alleviates the over-fitting and the degradation problem without model modification. Furthermore, PRUNE-TUNE is able to sequentially learn a single network with multiple disjoint domainspecific sub-networks for multiple domains. Empirical experiment results show that PRUNE-TUNE outperforms several strong competitors in the target domain test set without sacrificing the quality on the general domain in both single and multi-domain settings. The source code and data are available at https: //github. com/ohlionel/Prune-Tune.

PDF Details

AAAI Conference 2021 Conference Paper

Listen, Understand and Translate: Triple Supervision Decouples End-to-end Speech-to-text Translation

Qianqian Dong
Rong Ye
Mingxuan Wang
Hao Zhou
Shuang Xu
Bo Xu
Lei Li

An end-to-end speech-to-text translation (ST) takes audio in a source language and outputs the text in a target language. Existing methods are limited by the amount of parallel corpus. Can we build a system to fully utilize signals in a parallel ST corpus? We are inspired by human understanding system which is composed of auditory perception and cognitive processing. In this paper, we propose Listen-Understand- Translate, (LUT), a unified framework with triple supervision signals to decouple the end-to-end speech-to-text translation task. LUT is able to guide the acoustic encoder to extract as much information from the auditory input. In addition, LUT utilizes a pre-trained BERT model to enforce the upper encoder to produce as much semantic information as possible, without extra data. We perform experiments on a diverse set of speech translation benchmarks, including Librispeech English-French, IWSLT English-German and TED English-Chinese. Our results demonstrate LUT achieves the state-of-the-art performance, outperforming previous methods. The code is available at https: //github. com/dqqcasia/st.

PDF Details

AAAI Conference 2020 Conference Paper

Towards Making the Most of BERT in Neural Machine Translation

Jiacheng Yang
Mingxuan Wang
Hao Zhou
Chengqi Zhao
Weinan Zhang
Yong Yu
Lei Li

GPT-2 and BERT demonstrate the effectiveness of using pretrained language models (LMs) on various natural language processing tasks. However, LM ﬁne-tuning often suffers from catastrophic forgetting when applied to resource-rich tasks. In this work, we introduce a concerted training framework (CTNMT) that is the key to integrate the pre-trained LMs to neural machine translation (NMT). Our proposed CTNMT consists of three techniques: a) asymptotic distillation to ensure that the NMT model can retain the previous pre-trained knowledge; b) a dynamic switching gate to avoid catastrophic forgetting of pre-trained knowledge; and c) a strategy to adjust the learning paces according to a scheduled policy. Our experiments in machine translation show CTNMT gains of up to 3 BLEU score on the WMT14 English-German language pair which even surpasses the previous state-of-the-art pretraining aided NMT by 1. 4 BLEU score. While for the large WMT14 English-French task with 40 millions of sentencepairs, our base model still signiﬁcantly improves upon the state-of-the-art Transformer big model by more than 1 BLEU score.

PDF Details

AAAI Conference 2018 Conference Paper

Deep Semantic Role Labeling With Self-Attention

Zhixing Tan
Mingxuan Wang
Jun Xie
Yidong Chen
Xiaodong Shi

Semantic Role Labeling (SRL) is believed to be a crucial step towards natural language understanding and has been widely studied. Recent years, end-to-end SRL with recurrent neural networks (RNN) has gained increasing attention. However, it remains a major challenge for RNNs to handle structural information and long range dependencies. In this paper, we present a simple and effective architecture for SRL which aims to address these problems. Our model is based on self-attention which can directly capture the relationships between two tokens regardless of their distance. Our single model achieves F1 = 83. 4 on the CoNLL-2005 shared task dataset and F1 = 82. 7 on the CoNLL-2012 shared task dataset, which outperforms the previous state-of-the-art results by 1. 8 and 1. 0 F1 score respectively. Besides, our model is computationally efﬁcient, and the parsing speed is 50K tokens per second on a single Titan X GPU.

PDF Details

IJCAI Conference 2015 Conference Paper

Syntax-Based Deep Matching of Short Texts

Mingxuan Wang
Zhengdong Lu
Hang Li
Qun Liu

Many tasks in natural language processing, ranging from machine translation to question answering, can be reduced to the problem of matching two sentences or more generally two short texts. We propose a new approach to the problem, called Deep Match Tree (DEEPMATCHtree), under a general setting. The approach consists of two components, 1) a mining algorithm to discover patterns for matching two short-texts, defined in the product space of dependency trees, and 2) a deep neural network for matching short texts using the mined patterns, as well as a learning algorithm to build the network having a sparse structure. We test our algorithm on the problem of matching a tweet and a response in social media, a hard matching problem proposed in [Wang et al. , 2013], and show that DEEPMATCHtree can outperform a number of competitor models including one without using dependency trees and one based on word-embedding, all with large margins.

PDF Details