1 citations · 3 across the 32 of their papers we have counts for
32 papers
From Scoring to Acting: Outcome-Verified Comparative Self-Distillation for LLM Agents
Xu Xia, Jinghua Piao, Min Yang +3
Recent work on LLM agents is shifting from external capability elicitation to capability internalization, enabling agents to retain useful skills without retrieval at inference tim…
A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM Integration into Upcycled MoE
Hao Zhou, Tianhao Li, Zhijun Wang +6
Expanding Large Language Models~(LLMs) to new languages is a costly endeavor, demanding extensive Continued Pre-Training~(CPT) and data-intensive alignment. While recent data-free…
SkillMaster: Toward Autonomous Skill Mastery in LLM Agents
Min Yang, Jinghua Piao, Xu Xia +4
Skills provide an effective mechanism for improving LLM agents on complex tasks, yet in existing agent frameworks, their creation, refinement, and selection are typically governed…
TAPO: Translation Augmented Policy Optimization for Multilingual Mathematical Reasoning
Xu Huang, Zhejian Lai, Zixian Huang +2
Large Language Models (LLMs) have demonstrated remarkable proficiency in English mathematical reasoning, yet a significant performance disparity persists in multilingual contexts,…
ExpLang: Improved Exploration and Exploitation in LLM Reasoning with On-Policy Thinking Language Selection
Changjiang Gao, Zixian Huang, Kaichen Yang +3
Current large reasoning models (LRMs) have shown strong ability on challenging tasks after reinforcement learning (RL) based post-training. However, previous work mainly focuses on…
EnAnchored-X2X: English-Anchored Optimization for Many-to-Many Translation
Sen Yang, Yu Bao, Yu Lu +3
Large language models (LLMs) have demonstrated strong machine translation capabilities for English-centric language pairs but underperform in direct non-English (x2x) translation.…
SDGO: Self-Discrimination-Guided Optimization for Consistent Safety in Large Language Models
Peng Ding, Wen Sun, Dailin Li +4
Large Language Models (LLMs) excel at various natural language processing tasks but remain vulnerable to jailbreaking attacks that induce harmful content generation. In this paper,…
PATS: Process-Level Adaptive Thinking Mode Switching
Yi Wang, Junxiao Liu, Shimao Zhang +2
Current large-language models (LLMs) typically adopt a fixed reasoning strategy, either simple or complex, for all questions, regardless of their difficulty. This neglect of variat…
Why Not Act on What You Know? Unleashing Safety Potential of LLMs via Self-Aware Guard Enhancement
Peng Ding, Jun Kuang, Zongyu Wang +4
Large Language Models (LLMs) have shown impressive capabilities across various tasks but remain vulnerable to meticulously crafted jailbreak attacks. In this paper, we identify a c…
Trans-Zero: Self-Play Incentivizes Large Language Models for Multilingual Translation Without Parallel Data
Wei Zou, Sen Yang, Yu Bao +3
The rise of Large Language Models (LLMs) has reshaped machine translation (MT), but multilingual MT still relies heavily on parallel data for supervised fine-tuning (SFT), facing c…
How Does Alignment Enhance LLMs' Multilingual Capabilities? A Language Neurons Perspective
Shimao Zhang, Zhejian Lai, Xiang Liu +5
Multilingual Alignment is an effective and representative paradigm to enhance LLMs' multilingual capabilities, which transfers the capabilities from the high-resource languages to…
The First Impression Problem: Internal Bias Triggers Overthinking in Reasoning Models
Renfei Dang, Zhening Li, Shujian Huang +1
Reasoning models often exhibit overthinking, characterized by redundant reasoning steps. We identify \emph{internal bias} elicited by the input question as a key trigger of such be…
Understanding LLMs' Cross-Lingual Context Retrieval: How Good It Is And Where It Comes From
Changjiang Gao, Hankun Lin, Xin Huang +5
Cross-lingual context retrieval (extracting contextual information in one language based on requests in another) is a fundamental aspect of cross-lingual alignment, but the perform…
R-PRM: Reasoning-Driven Process Reward Modeling
Shuaijie She, Junxiao Liu, Yifeng Liu +3
Large language models (LLMs) inevitably make mistakes when performing step-by-step mathematical reasoning. Process Reward Models (PRMs) have emerged as a promising solution by eval…
Generalizing From Short to Long: Effective Data Synthesis for Long-Context Instruction Tuning
Wenhao Zhu, Pinzhen Chen, Hanxu Hu +4
Long-context modelling for large language models (LLMs) has been a key area of recent research because many real world use cases require reasoning over longer inputs such as docume…
Alleviating Distribution Shift in Synthetic Data for Machine Translation Quality Estimation
Xiang Geng, Zhejian Lai, Jiajun Chen +2
Quality Estimation (QE) models evaluate the quality of machine translations without reference translations, serving as the reward models for the translation task. Due to the data s…
Extend Adversarial Policy Against Neural Machine Translation via Unknown Token
Wei Zou, Shujian Huang, Jiajun Chen
Generating adversarial examples contributes to mainstream neural machine translation~(NMT) robustness. However, popular adversarial policies are apt for fixed tokenization, hinderi…
Formality is Favored: Unraveling the Learning Preferences of Large Language Models on Data with Conflicting Knowledge
Jiahuan Li, Yiqing Cao, Shujian Huang +1
Having been trained on massive pretraining data, large language models have shown excellent performance on many knowledge-intensive tasks. However, pretraining data tends to contai…
MoE-LPR: Multilingual Extension of Large Language Models through Mixture-of-Experts with Language Priors Routing
Hao Zhou, Zhijun Wang, Shujian Huang +6
Large Language Models (LLMs) are often English-centric due to the disproportionate distribution of languages in their pre-training data. Enhancing non-English language capabilities…
Hallu-PI: Evaluating Hallucination in Multi-modal Large Language Models within Perturbed Inputs
Peng Ding, Jingyu Wu, Jun Kuang +6
Multi-modal Large Language Models (MLLMs) have demonstrated remarkable performance on various visual-language understanding and generation tasks. However, MLLMs occasionally genera…
Multilingual Contrastive Decoding via Language-Agnostic Layers Skipping
Wenhao Zhu, Sizhe Liu, Shujian Huang +3
Decoding by contrasting layers (DoLa), is designed to improve the generation quality of large language models (LLMs) by contrasting the prediction probabilities between an early ex…
PreAlign: Boosting Cross-Lingual Transfer by Early Establishment of Multilingual Alignment
Jiahuan Li, Shujian Huang, Aarron Ching +2
Large language models demonstrate reasonable multilingual abilities, despite predominantly English-centric pretraining. However, the spontaneous multilingual alignment in these mod…
Getting More from Less: Large Language Models are Good Spontaneous Multilingual Learners
Shimao Zhang, Changjiang Gao, Wenhao Zhu +6
Recently, Large Language Models (LLMs) have shown impressive language capabilities. While most of the existing LLMs have very unbalanced performance across different languages, mul…
Large Language Models are Limited in Out-of-Context Knowledge Reasoning
Peng Hu, Changjiang Gao, Ruiqi Gao +2
Large Language Models (LLMs) possess extensive knowledge and strong capabilities in performing in-context reasoning. However, previous work challenges their out-of-context reasonin…