Research map / Continual learning / Continual fine-tuning of LLMs
Can BERT Refrain from Forgetting on Sequential Tasks? A Probing Study
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Train sequentially without memory replay, then re-train task-specific linear probes while freezing the encoder | BERT_BASE · Text classification, five tasks | macro-averaged accuracy: 75.87%±0.73%, 76.76%±0.64%, 75.19%±0.43%, 76.76%±0.71% | Helped |
| Train sequentially without any memory replay | BERT_BASE · Extractive QA tasks | F1 | No clear effect |
| Increase the replay interval | BERT task-incremental learning · Text classification, previously learned tasks | probing accuracy | Hurt |
| Add sparse memory replay of prior-task examples | BERT task-incremental learning · Text classification, inter-task representations | class/task separability | Helped |
| Use a median-sized neighbor set to measure topological ordering (ablation) | AGNews class representations after learning Amazon · AGNews | Pearson correlation coefficient: 81.09 to 81.68; 48.35 to 51.46; 83.11 to 83.90; 72.41 to 73.98 | Mixed |
Closest papers by meaning
- Spurious Forgetting in Continual Learning of Language Models (ICLR 2025 · similarity 0.88)
- Demystifying Language Model Forgetting with Low-rank Example Associations (NeurIPS 2025 · similarity 0.88)
- Learn more, but bother less: parameter efficient continual learning (NeurIPS 2024 · similarity 0.87)
- Kolmogorov-Arnold Networks Still Catastrophically Forget but Differently from MLP (AAAI 2025 · similarity 0.87)
- Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective (ICML 2025 · similarity 0.86)
- Predicting the Susceptibility of Examples to Catastrophic Forgetting (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.