no code implementations • 16 Feb 2024 • Shengjie Qiu, Junhao Zheng, Zhen Liu, Yicheng Luo, Qianli Ma
As for the E2O problem, we use knowledge distillation to maintain the model's discriminative ability for old entities.
1 code implementation • 13 Feb 2024 • Junhao Zheng, Shengjie Qiu, Qianli Ma
However, existing IL scenarios and datasets are unqualified for assessing forgetting in PLMs, giving an illusion that PLMs do not suffer from catastrophic forgetting.
1 code implementation • 13 Dec 2023 • Junhao Zheng, Shengjie Qiu, Qianli Ma
Most assume that catastrophic forgetting is the biggest obstacle to achieving superior IL performance and propose various techniques to overcome this issue.
1 code implementation • 19 Jun 2023 • Junhao Zheng, Qianli Ma, Shengjie Qiu, Yue Wu, Peitian Ma, Junlong Liu, Huawen Feng, Xichen Shang, Haibin Chen
Intriguingly, the unified objective can be seen as the sum of the vanilla fine-tuning objective, which learns new knowledge from target data, and the causal objective, which preserves old knowledge from PLMs.