---
title: "TeleAI at SemEval-2026 Task 3 — paper by Zhou et al. (2026)"
description: "This paper describes TeleAI's entry for Track A, Subtask 1 (DimASR) of SemEval-2026 Task 3. 28 references, cited by 1 paper here."
authors: ["Yan Zhou", "Wangshicheng Wang", "Shiquan Wang", "Mengjiao Bao", "Ruiyu Fang", "Shuangyong Song", "Yongxiang Li", "Xuelong Li"]
year: 2026
venue: "Proceedings of the 20th International Workshop on Semantic Evaluation (2026)"
source: "https://aclanthology.org/2026.semeval-1.233/"
doi: "10.18653/v1/2026.semeval-1.233"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/zhou-2026-teleai-semeval-2026-task-3/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhou-2026-teleai-semeval-2026-task-3/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/18/) / TeleAI at SemEval-2026 Task 3: Large Language Models for Dimensional Aspect-Based Sentiment Analysis

# TeleAI at SemEval-2026 Task 3: Large Language Models for Dimensional Aspect-Based Sentiment Analysis

**Authors:** [Yan Zhou](https://md.research.elamur.ai/free/en/authors/yan-zhou/), Wangshicheng Wang, [Shiquan Wang](https://md.research.elamur.ai/free/en/authors/shiquan-wang/), [Mengjiao Bao](https://md.research.elamur.ai/free/en/authors/mengjiao-bao/), [Ruiyu Fang](https://md.research.elamur.ai/free/en/authors/ruiyu-fang/), [Shuangyong Song](https://md.research.elamur.ai/free/en/authors/shuangyong-song/), [Yongxiang Li](https://md.research.elamur.ai/free/en/authors/yongxiang-li/) and [Xuelong Li](https://md.research.elamur.ai/free/en/authors/xuelong-li/) · **Year:** 2026 · **Venue:** [Proceedings of the 20th International Workshop on Semantic Evaluation (2026)](https://md.research.elamur.ai/free/en/venues/semeval/) ([SemEval 2026](https://md.research.elamur.ai/free/en/venues/semeval/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.semeval-1.233/) · [PDF](https://aclanthology.org/2026.semeval-1.233.pdf) · [DOI 10.18653/v1/2026.semeval-1.233](https://doi.org/10.18653/v1/2026.semeval-1.233) · [OpenAlex](https://openalex.org/W7166806618)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read

## Summary
This paper describes TeleAI's entry for Track A, Subtask 1 (DimASR) of SemEval-2026 Task 3. In this task, sentiment toward each aspect term is not a polarity label but a pair of continuous Valence–Arousal (VA) scores between 1 and 9, scored with RMSE_VA. The authors treat the problem as direct regression. The input joins the aspect and the text using [ASPECT] and [TEXT] markers. Qwen2.5-7B encodes it, mask-aware mean pooling summarizes the hidden states, and a linear head outputs two values, which a sigmoid squashes into the valid range. Only LoRA adapters on the attention projections and the head are trained. Training pools all language–domain training files and adds Huber loss, R-Drop-style consistency, and embedding-level PGD adversarial perturbations. A per-dimension linear calibrator, fitted on a local validation split, adjusts the outputs. On the English-Restaurant dev phase, the system reached RMSE_VA 0.85 and placed first on the leaderboard. The work shows that a mid-sized LLM with parameter-efficient tuning and regularization can beat both full fine-tuning and larger models when labeled data is scarce.

## Key points
- On English-Restaurant dev, the best system scores 0.85 RMSE_VA. A Qwen2.5-7B baseline with full SFT and plain MSE scores 1.03.
- The ablations identify PGD and R-Drop as the most important components. Removing them raises error to 0.94 and 0.92. Switching LoRA (r=16, α=32) to full fine-tuning gives 0.91.
- Qwen2.5-7B outperformed Qwen2.5-14B and Qwen2.5-72B, which the authors attribute to overfitting on limited data. Qwen models also outperformed BERT-style encoders.
- Two ideas brought no consistent gain and were dropped: multi-stage domain transfer and a negative Pearson correlation loss. Extending max_len from 128 to 256 also did not help much.
- The main remaining errors are:
  - arousal predictions drifting toward the middle of the scale;
  - sentiment spilling across aspects in contrastive sentences;
  - truncation cutting off key evidence.
- Simple JSON prompting for Subtasks 2–3 was weak. Code is released at github.com/yanzhou06/SemEval-2026-Task3-subtask1-code.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [China Telecom](https://md.research.elamur.ai/free/en/companies/china-telecom-cn/) — Company, Beijing, China · 9 works · 16 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,649 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 930 works

🧭 Topics (2): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,894 works · [😊 Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · 1,035 works

## Cite

Cite as: Yan Zhou, Wangshicheng Wang, Shiquan Wang, Mengjiao Bao, Ruiyu Fang, Shuangyong Song, Yongxiang Li and Xuelong Li. 2026. TeleAI at SemEval-2026 Task 3: Large Language Models for Dimensional Aspect-Based Sentiment Analysis. Proceedings of the 20th International Workshop on Semantic Evaluation (2026).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{zhou2026teleai,
  title = {{TeleAI at SemEval-2026 Task 3: Large Language Models for Dimensional Aspect-Based Sentiment Analysis}},
  author = {Yan Zhou, Wangshicheng Wang, Shiquan Wang, Mengjiao Bao, Ruiyu Fang, Shuangyong Song, Yongxiang Li and Xuelong Li},
  booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (2026)},
  year = {2026},
  doi = {10.18653/v1/2026.semeval-1.233},
  url = {https://doi.org/10.18653/v1/2026.semeval-1.233}
}
```

- [← previous TeamV at LT-EDI 2026: Multilingual Hate Speech Span Detection and Counter-Narrative Generation via Few-Shot In-Context Learning](https://md.research.elamur.ai/free/en/papers/ulli-2026-teamv-lt-edi-2026-multilingual/)
- [next → TELL-TALE: Task Efficient LLMs with Task Aware Layer Elimination](https://md.research.elamur.ai/free/en/papers/naim-2026-tell-tale-task-efficient-llms/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/18/)

## References

28 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Hongjun An, Wenhan Hu, Sida Huang, Siqi Huang, Ruanjun Li, Yuanzhi Liang, Jiawei Shao, Yiliang Song, Zihan Wang, Cheng Yuan, Chi Zhang, Hongyuan Zhang, Wenhao Zhuang, and Xuelong Li. 2026. AI flow: perspectives, scenarios, and approaches. Vicinagearth, 3:1. [find](https://scholar.google.com/scholar?q=Hongjun+An%2C+Wenhan+Hu%2C+Sida+Huang%2C+Siqi+Huang%2C+Ruanjun+Li%2C+Yuanzhi+Liang%2C+Jiawei+Shao%2C+Yiliang+Song%2C+Zihan+Wang%2C+Cheng+Yuan%2C+Chi+Zhang%2C+Hongyuan+Zhang%2C+Wenhao+Zhuang%2C+and+Xuelong+Li%2E+2026%2E+AI+flow%3A+pe)
2. Jinyang Chen, Haolun Wu, Jianhong Pang, Yihua Wang, Dell Zhang, and Changzhi Sun. 2025. Tool learning with language models: a comprehensive survey of methods, pipelines, and benchmarks. Vicinagearth, 2:16. [find](https://scholar.google.com/scholar?q=Jinyang+Chen%2C+Haolun+Wu%2C+Jianhong+Pang%2C+Yihua+Wang%2C+Dell+Zhang%2C+and+Changzhi+Sun%2E+2025%2E+Tool+learning+with+language+models%3A+a+comprehensive+survey+of+methods%2C+pipelines%2C+and+benchmarks%2E+Vicinagearth%2C+)
3. Zhongjiang He, Zihan Wang, Xinzhang Liu, Shixuan Liu, Yitong Yao, Yuyao Huang, Xuelong Li, Yongxiang Li, Zhonghao Che, Zhaoxi Zhang, Yan Wang, Xin Wang, Luwen Pu, Huinan Xu, Ruiyu Fang, Yu Zhao, Jie Zhang, Xiaomeng Huang, Zhilong Lu, and 17 others. 2024. [TeleChat technical report](https://md.research.elamur.ai/free/en/papers/he-2024-telechat-technical-report/). Preprint, arXiv:24…
4. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, and Weizhu Chen. 2021. [Lora: Low-rank adaptation of large language models](https://md.research.elamur.ai/free/en/papers/hu-2021-lora-low-rank-adaptation-large/). Preprint, arXiv:2106.09685.
5. Bin-Bin Jia and Min-Ling Zhang. 2024. Multidimensional classification: paradigm, algorithms and beyond. Vicinagearth, 1:3. [find](https://scholar.google.com/scholar?q=Bin%2DBin+Jia+and+Min%2DLing+Zhang%2E+2024%2E+Multidimensional+classification%3A+paradigm%2C+algorithms+and+beyond%2E+Vicinagearth%2C+1%3A3%2E)
6. Lung-Hao Lee, Liang-Chih Yu, Natalia Loukashe-vich, Ilseyar Alimova, Alexander Panchenko, Tzu-Mi Lin, Zhe-Yu Xu, Jian-Yu Zhou, Guangmin Zheng, Jin Wang, Sharanya Awasthi, Jonas Becker, Jan Philip Wahle, Terry Ruas, Shamsuddeen Hassan Muhammad, and Saif M. Mohammad. 2026. Dimabsa: Building multilingu… [find](https://scholar.google.com/scholar?q=Lung%2DHao+Lee%2C+Liang%2DChih+Yu%2C+Natalia+Loukashe%2Dvich%2C+Ilseyar+Alimova%2C+Alexander+Panchenko%2C+Tzu%2DMi+Lin%2C+Zhe%2DYu+Xu%2C+Jian%2DYu+Zhou%2C+Guangmin+Zheng%2C+Jin+Wang%2C+Sharanya+Awasthi%2C+Jonas+Becker%2C+Jan+Philip+Wahl)
7. Xiang Li, Yiqun Yao, Xin Jiang, Xuezhi Fang, Chao Wang, Xinzhang Liu, Zihan Wang, Yu Zhao, Xin Wang, Yuyao Huang, Shuangyong Song, Yongxiang Li, Zheng Zhang, Bo Zhao, Aixin Sun, Yequan Wang, Zhongjiang He, Zhongyuan Wang, Xuelong Li, and Tiejun Huang. 2024a. 52B to 1T: Lessons learned via Tele-FLM s… — [arXiv:2407.02783](https://arxiv.org/abs/2407.02783) [find](https://scholar.google.com/scholar?q=Xiang+Li%2C+Yiqun+Yao%2C+Xin+Jiang%2C+Xuezhi+Fang%2C+Chao+Wang%2C+Xinzhang+Liu%2C+Zihan+Wang%2C+Yu+Zhao%2C+Xin+Wang%2C+Yuyao+Huang%2C+Shuangyong+Song%2C+Yongxiang+Li%2C+Zheng+Zhang%2C+Bo+Zhao%2C+Aixin+Sun%2C+Yequan+Wang%2C+Zhongjian)
8. Xiang Li, Yiqun Yao, Xin Jiang, Xuezhi Fang, Chao Wang, Xinzhang Liu, Zihan Wang, Yu Zhao, Xin Wang, Yuyao Huang, Shuangyong Song, Yongxiang Li, Zheng Zhang, Bo Zhao, Aixin Sun, Yequan Wang, Zhongjiang He, Zhongyuan Wang, Xuelong Li, and Tiejun Huang. 2024b. Tele-FLM technical report. Preprint, arXi… — [arXiv:2404.16645](https://arxiv.org/abs/2404.16645) [find](https://scholar.google.com/scholar?q=Xiang+Li%2C+Yiqun+Yao%2C+Xin+Jiang%2C+Xuezhi+Fang%2C+Chao+Wang%2C+Xinzhang+Liu%2C+Zihan+Wang%2C+Yu+Zhao%2C+Xin+Wang%2C+Yuyao+Huang%2C+Shuangyong+Song%2C+Yongxiang+Li%2C+Zheng+Zhang%2C+Bo+Zhao%2C+Aixin+Sun%2C+Yequan+Wang%2C+Zhongjian)
9. Kaidong Yu, Yu Zhao, Ruiyu Fang, and 35 others. 2025. Training report of TeleChat3-MoE. Preprint, arXiv:2512.24157. — [arXiv:2512.24157](https://arxiv.org/abs/2512.24157) [find](https://scholar.google.com/scholar?q=Kaidong+Yu%2C+Yu+Zhao%2C+Ruiyu+Fang%2C+and+35+others%2E+2025%2E+Training+report+of+TeleChat3%2DMoE%2E+Preprint%2C+arXiv%3A2512%2E24157%2E+%E2%80%94+%5BarXiv%3A2512%2E24157%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2512%2E24157%29)
10. Yong Luo, Yonggang Wen, Han Hu, Bo Du, Ling-Yu Duan, and Dacheng Tao. 2024. Transfer metric learning: algorithms, applications and outlooks. Vicinagearth, 1:2. [find](https://scholar.google.com/scholar?q=Yong+Luo%2C+Yonggang+Wen%2C+Han+Hu%2C+Bo+Du%2C+Ling%2DYu+Duan%2C+and+Dacheng+Tao%2E+2024%2E+Transfer+metric+learning%3A+algorithms%2C+applications+and+outlooks%2E+Vicinagearth%2C+1%3A2%2E)
11. Aleksander Madry, Aleksandar Makelov, Ludwig Schmidt, Dimitris Tsipras, and Adrian Vladu. 2017. [Towards deep learning models resistant to adversarial attacks](https://md.research.elamur.ai/free/en/papers/madry-2017-towards-deep-learning-models-resistant/). Preprint, arXiv:1706.06083.
12. Saif M. Mohammad. 2018. [Obtaining reliable human ratings of valence, arousal, and dominance for 20,000 english words](https://md.research.elamur.ai/free/en/papers/mohammad-2018-obtaining-reliable-human-ratings-valence/). In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Association for Computational Linguistics.
13. PyTorch Contributors. Pytorch documentation: torch.nn.SmoothL1Loss. [https://pytorch.org/docs/stable/generated/torch](https://pytorch.org/docs/stable/generated/torch). nn.SmoothL1Loss.html. Accessed 2026-02-23. [↗](https://pytorch.org/docs/stable/generated/torch)
14. Qwen Team. 2025. Qwen2.5 technical report. Preprint, arXiv:2412.15115. [find](https://scholar.google.com/scholar?q=Qwen+Team%2E+2025%2E+Qwen2%2E5+technical+report%2E+Preprint%2C+arXiv%3A2412%2E15115%2E)
15. Shuangyong Song and Yao Meng. 2015a. Classifying and ranking microblogging hashtags with news categories. In 2015 IEEE 9th International Conference on Research Challenges in Information Science (RCIS), pages 540–541. [find](https://scholar.google.com/scholar?q=Shuangyong+Song+and+Yao+Meng%2E+2015a%2E+Classifying+and+ranking+microblogging+hashtags+with+news+categories%2E+In+2015+IEEE+9th+International+Conference+on+Research+Challenges+in+Information+Science+%28RCIS%29)
16. Shuangyong Song and Yao Meng. 2015b. Detecting concept-level emotion cause in microblogging. In Proceedings of the 24th International Conference on World Wide Web Companion, pages 119–120. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Shuangyong+Song+and+Yao+Meng%2E+2015b%2E+Detecting+concept%2Dlevel+emotion+cause+in+microblogging%2E+In+Proceedings+of+the+24th+International+Conference+on+World+Wide+Web+Companion%2C+pages+119%E2%80%93120%2E+Association)
17. Shuangyong Song, Chao Wang, Haiqing Chen, and Huan Chen. 2021a. An emotional comfort framework for improving user satisfaction in E-commerce customer service chatbots. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language… [find](https://scholar.google.com/scholar?q=Shuangyong+Song%2C+Chao+Wang%2C+Haiqing+Chen%2C+and+Huan+Chen%2E+2021a%2E+An+emotional+comfort+framework+for+improving+user+satisfaction+in+E%2Dcommerce+customer+service+chatbots%2E+In+Proceedings+of+the+2021+Confe)
18. Shuangyong Song, Chao Wang, Siyang Liu, Haiqing Chen, Huan Chen, and Hongyun Bao. 2020. Sentiment analysis technologies in AliMe—an intelligent assistant for E-commerce. International Journal of Asian Language Processing, 30(04):2050016. [find](https://scholar.google.com/scholar?q=Shuangyong+Song%2C+Chao+Wang%2C+Siyang+Liu%2C+Haiqing+Chen%2C+Huan+Chen%2C+and+Hongyun+Bao%2E+2020%2E+Sentiment+analysis+technologies+in+AliMe%E2%80%94an+intelligent+assistant+for+E%2Dcommerce%2E+International+Journal+of+Asian)
19. Xiaobo Liang, Lijun Liu, Zhiwei Liu, Yicheng He, and Liang Zhao. 2021. R-drop: Regularized dropout for neural networks. Preprint, arXiv:2106.14448. — [arXiv:2106.14448](https://arxiv.org/abs/2106.14448) [find](https://scholar.google.com/scholar?q=Xiaobo+Liang%2C+Lijun+Liu%2C+Zhiwei+Liu%2C+Yicheng+He%2C+and+Liang+Zhao%2E+2021%2E+R%2Ddrop%3A+Regularized+dropout+for+neural+networks%2E+Preprint%2C+arXiv%3A2106%2E14448%2E+%E2%80%94+%5BarXiv%3A2106%2E14448%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2106%2E1444)
20. Shuangyong Song, Kexin Wang, Chao Wang, Haiqing Chen, and Huan Chen. 2021b. An emotion-controlled dialog response generation model with dynamic vocabulary. Preprint, arXiv:2103.02878. — [arXiv:2103.02878](https://arxiv.org/abs/2103.02878) [find](https://scholar.google.com/scholar?q=Shuangyong+Song%2C+Kexin+Wang%2C+Chao+Wang%2C+Haiqing+Chen%2C+and+Huan+Chen%2E+2021b%2E+An+emotion%2Dcontrolled+dialog+response+generation+model+with+dynamic+vocabulary%2E+Preprint%2C+arXiv%3A2103%2E02878%2E+%E2%80%94+%5BarXiv%3A2103%2E02)
21. Xinzhang Liu, Chao Wang, Zhihao Yang, Zhuo Jiang, Xuncheng Zhao, Haoran Wang, Lei Li, Dongdong He, Luobin Liu, Kaizhe Yuan, Han Gao, Zihan Wang, Yitong Yao, Sishi Xiong, Wenmin Deng, Haowei He, Shiquan Wang, Ruiyu Fang, Zhongjiang He, Shuangyong Song, and Yongxiang Li. 2025a. Emotional support with… — [arXiv:2507.12820](https://arxiv.org/abs/2507.12820) [find](https://scholar.google.com/scholar?q=Xinzhang+Liu%2C+Chao+Wang%2C+Zhihao+Yang%2C+Zhuo+Jiang%2C+Xuncheng+Zhao%2C+Haoran+Wang%2C+Lei+Li%2C+Dongdong+He%2C+Luobin+Liu%2C+Kaizhe+Yuan%2C+Han+Gao%2C+Zihan+Wang%2C+Yitong+Yao%2C+Sishi+Xiong%2C+Wenmin+Deng%2C+Haowei+He%2C+Shiqua)
22. Shiquan Wang, Ruiyu Fang, Mengxiang Li, Zhongjiang He, and Shuangyong Song. 2025b. Empathetic dialogue generation with LLMs for emotional support. In Natural Language Processing and Chinese Computing, volume 16105 of Lecture Notes in Computer Science, pages 533–540. Springer. [find](https://scholar.google.com/scholar?q=Shiquan+Wang%2C+Ruiyu+Fang%2C+Mengxiang+Li%2C+Zhongjiang+He%2C+and+Shuangyong+Song%2E+2025b%2E+Empathetic+dialogue+generation+with+LLMs+for+emotional+support%2E+In+Natural+Language+Processing+and+Chinese+Computing%2C)
23. Shiquan Wang, Mengxiang Li, Shengxiong Peng, Fang Yu, Zhongjiang He, Shuangyong Song, and Yongxiang Li. 2025c. TeleAI at SemEval-2025 task 11: Bridging the gap in text-based emotion detection with prompt engineering and data augmentation. In Proceedings of the 19th International Workshop on Semantic… [find](https://scholar.google.com/scholar?q=Shiquan+Wang%2C+Mengxiang+Li%2C+Shengxiong+Peng%2C+Fang+Yu%2C+Zhongjiang+He%2C+Shuangyong+Song%2C+and+Yongxiang+Li%2E+2025c%2E+TeleAI+at+SemEval%2D2025+task+11%3A+Bridging+the+gap+in+text%2Dbased+emotion+detection+with+pro)
24. Zihan Wang, XinZhang Liu, Shixuan Liu, Yitong Yao, Yunyao Huang, Mengxiang Li, Zhongjiang He, Yongxian Li, Luwen Pu, Huinan Xu, Chao Wang, and Shuangyong Song. 2024. TeleChat: An open-source billingual large language model. In Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (S… [find](https://scholar.google.com/scholar?q=Zihan+Wang%2C+XinZhang+Liu%2C+Shixuan+Liu%2C+Yitong+Yao%2C+Yunyao+Huang%2C+Mengxiang+Li%2C+Zhongjiang+He%2C+Yongxian+Li%2C+Luwen+Pu%2C+Huinan+Xu%2C+Chao+Wang%2C+and+Shuangyong+Song%2E+2024%2E+TeleChat%3A+An+open%2Dsource+billingua)
25. Zihan Wang, Xinzhang Liu, Yitong Yao, Chao Wang, Yu Zhao, Zhihao Yang, Wenmin Deng, Kaipeng Jia, Jiaxin Peng, Yuyao Huang, Sishi Xiong, Zhuo Jiang, Kaidong Yu, Xiaohui Hu, Fubei Yao, Ruiyu Fang, Zhuoru Jiang, Ruiting Song, Qiyi Xie, and 19 others. 2025d. Technical report of TeleChat2, TeleChat2.5 an… — [arXiv:2507.18013](https://arxiv.org/abs/2507.18013) [find](https://scholar.google.com/scholar?q=Zihan+Wang%2C+Xinzhang+Liu%2C+Yitong+Yao%2C+Chao+Wang%2C+Yu+Zhao%2C+Zhihao+Yang%2C+Wenmin+Deng%2C+Kaipeng+Jia%2C+Jiaxin+Peng%2C+Yuyao+Huang%2C+Sishi+Xiong%2C+Zhuo+Jiang%2C+Kaidong+Yu%2C+Xiaohui+Hu%2C+Fubei+Yao%2C+Ruiyu+Fang%2C+Zhuor)
26. Liang-Chih Yu, Jonas Becker, Shamsuddeen Hassan Muhammad, Idris Abdulmumin, Lung-Hao Lee, Ying-Lung Lin, Jin Wang, Jan Philip Wahle, Terry Ruas, Alexander Panchenko, Ilseyar Alimova, Kai-Wei Chang, Lilian Wanzare, Nelson Odhiambo, Bela Gipp, and Saif M. Mohammad. 2026. SemEval-2026 task 3: Dimension… [find](https://scholar.google.com/scholar?q=Liang%2DChih+Yu%2C+Jonas+Becker%2C+Shamsuddeen+Hassan+Muhammad%2C+Idris+Abdulmumin%2C+Lung%2DHao+Lee%2C+Ying%2DLung+Lin%2C+Jin+Wang%2C+Jan+Philip+Wahle%2C+Terry+Ruas%2C+Alexander+Panchenko%2C+Ilseyar+Alimova%2C+Kai%2DWei+Chang%2C+Li)
27. Deji Zhao, Donghong Han, Ye Yuan, Bo Ning, Yongxiang Li, Zhongjiang He, Chao Wang, and Shuangyong Song. 2026. MEGE: A mixed emotion graph model for empathetic dialogue generation. Neural Networks, 195:108252. [find](https://scholar.google.com/scholar?q=Deji+Zhao%2C+Donghong+Han%2C+Ye+Yuan%2C+Bo+Ning%2C+Yongxiang+Li%2C+Zhongjiang+He%2C+Chao+Wang%2C+and+Shuangyong+Song%2E+2026%2E+MEGE%3A+A+mixed+emotion+graph+model+for+empathetic+dialogue+generation%2E+Neural+Networks%2C+195)
28. Hyperparameter Value / Setting Epochs 10 (pretrain_ratio=1.0 at best config) Max Sequence Length 128 Batch Size per_gpu=4, grad_accum=1 Optimizer AdamW, λ wd = 0.01 Learning Rate η head = 10 −3 ; η LoRA = 10 −4 ; warmup 0.1 + linear decay Precision bf16 (grad clip=1.0, seed=42) LoRA Config r=16, α=3… [find](https://scholar.google.com/scholar?q=Hyperparameter+Value+%2F+Setting+Epochs+10+%28pretrain%5Fratio%3D1%2E0+at+best+config%29+Max+Sequence+Length+128+Batch+Size+per%5Fgpu%3D4%2C+grad%5Faccum%3D1+Optimizer+AdamW%2C+%CE%BB+wd+%3D+0%2E01+Learning+Rate+%CE%B7+head+%3D+10+%E2%88%923+%3B+%CE%B7+Lo)

*Updated 2 November 2026*