Publications

5 results for Changsheng Wang

Reasoning Model Unlearning: Forgetting Traces, Not Just Answers, While Preserving Reasoning Skills
- - Changsheng Wang
  - Chongyu Fan
  - et al.
- 2025
- EMNLP 2025
Conference paper
LLM Unlearning on Noisy Forget Sets: A Study of Incomplete, Rewritten, and Watermarked Data
- - Changsheng Wang
  - Yihua Zhang
  - et al.
- 2025
- AISec 2025
Workshop paper
Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-Tuning
- - Changsheng Wang
  - Yihua Zhang
  - et al.
- 2025
- ICML 2025
Conference paper
Noisy But Forgotten: LLM Unlearning are Robust against Perturbed Data in the Wild
- - Changsheng Wang
  - Yihua Zhang
  - et al.
- 2025
- ICML 2025
Workshop paper
Rethinking Unlearning for Large Reasoning Models Download PDF
- - Changsheng Wang
  - Chongyu Fan
  - et al.
- 2025
- ICML 2025
Workshop paper