About Me

I am interning for Long Reasoning Models in the Tencent AI Lab.
I am also a Ph.D. candidate at NTU supervised by Prof. Chen Lihui.
I obtained my MSc Degree in Signal Processing at the same university. Before that, I obtained my Bachelor’s degree at UESTC.

Projects

Scaling RL-Zero: Towards more general and frontier reasoning tasks.
TransAgents: A virtual multi-agent translation company that mirrors the traditional translation publication processes.Paper.Code.
Complex instruction following: Project for Hunyuan LLMs.
Seamless speech-to-speech translation: Speach-to-speech translation model for Tencent Meeting App.
Privacy-preserving NMT system: A NMT translation system, which auto-protects the privacy information.
Back Translation NMT system: A back translation data algorithm to enhance the performance of the Tencent Translator App.

* denotes co-first authors.

Li, Y.*, Xu, J.*, Liang, T.*, Chen, X., He, Z., Liu, Q., Wang, R., Zhang, Z., Tu, Z., Mi, H., and Yu, D. Dancing with Critiques: Enhancing LLM Reasoning with Stepwise Natural Language Self-Critique.
Ji, K.*, Xu, J.*, Liang, T.*, Liu, Q.*, He, Z., Chen, X., Liu, X., Wang, Z., Chen, J., Wang, B. and Tu, Z., The First Few Tokens Are All You Need: An Efficient and Effective Unsupervised Prefix Fine-Tuning Method for Reasoning Models.
Wang, Y.*, Liu, Q.*, Xu, J.*, Liang, T.*, Chen, X., He, Z., Song, L., Yu, D., Li, J., Zhang, Z. and Wang, R., 2025. Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs. arXiv preprint arXiv:2501.18585.
Chen, X.*, Xu, J.*, Liang, T.*, He, Z.*, Pang, J., Yu, D., Song, L., Liu, Q., Zhou, M., Zhang, Z. and Wang, R., 2024. Do not think that much for 2+ 3=? on the overthinking of o1-like llms. ICML2025
Yu, D., Zhang, Y., Xu, J., Liang, T., Song, L., Tu, Z., Mi, H. and Yu, D., 2024. Teaching LLMs to Refine with Tools. arXiv preprint arXiv:2412.16871
Lin, Z.*, Liang, T.*, Xu, J.*, Wang, X., Luo, R., Shi, C., Li, S., Yang, Y. and Tu, Z., 2024. Critical Tokens Matter: Token-Level Contrastive Estimation Enhence LLM’s Reasoning Capability. ICML2025

Zhang, Z., Xu, J., Liang, T., Chen, X., He, Z., Wang, R. and Tu, Z., 2024. Draft Model Knows When to Stop: A Self-Verification Length Policy for Speculative Decoding. arXiv preprint arXiv:2411.18462.

Xu, J., Zhanyi, C.S., Xu, L. and Chen, L., 2024. Blendcse: Blend contrastive learnings for sentence embeddings with rich semantics and transferability. Expert Systems with Applications, 238, p.121909.
Xu, J., Shao, W., Chen, L. and Liu, L., 2023, December. DistillCSE: Distilled Contrastive Learning for Sentence Embeddings. In Findings of the Association for Computational Linguistics: EMNLP 2023 (pp. 8153-8165).
Xu, J., Shao, W., Chen, L. and Liu, L., 2023, December. SimCSE++: Improving Contrastive Learning for Sentence Embeddings from Two Perspectives. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (pp. 12028-12040).

Xu, J., Ruan, Y., Bi, W., Huang, G., Shi, S., Chen, L. and Liu, L., 2022, July. On Synthetic Data for Back Translation. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (pp. 419-430).

I would like to express my heartfelt gratitude to Principal Researcher Zhaopeng Tu and Lemao Liu for their invaluable guidance and support during my internship at Tencent AI Lab. Their expertise, encouragement, and insightful feedback have been instrumental in shaping my understanding of long-reasoning models and advancing my research skills.
I would also like to extend my thanks to the interns who worked alongside me during this journey. Their collaboration, diverse perspectives, and shared enthusiasm for innovation greatly enriched my experience and made our work environment both dynamic and inspiring.