Research map / Continual learning / Continual fine-tuning of LLMs
OnEDIT: Online Editing with Decoupled Implicit Task for Large Language Models
Chae-Won Lee, Jae-Hong Lee, Ji-Hun Kang, Joon-Hyuk Chang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| use OnEDIT instead of GainLoRA | T5-large · Long Sequence benchmark | AP: 29.8 -> 38.4 | Improved |
| use OnEDIT instead of GainLoRA | T5-large · Standard CIT benchmark | AP: 66.4 -> 67.4 | Improved |
| use OnEDIT instead of GainLoRA | T5-large · SuperNI benchmark | AP: 27.6 -> 28.9 | Improved |
| use OnEDIT instead of GainLoRA | T5-large · Standard CIT benchmark | forgetting rate: 24.8 -> 22.3 | Improved |
| use OnEDIT instead of GainLoRA | T5-large · SuperNI benchmark | forgetting rate: 12.3 -> 15.4 | Hurt |
| use OnEDIT instead of plain LoRA | T5-large · Long Sequence benchmark | AP: 17.8 -> 38.4 | Improved |
| use OnEDIT instead of GainLoRA | Llama3-8B · Standard CIT benchmark | AP: 67.4 -> 69.6 | Improved |
| use OnEDIT instead of GainLoRA | Llama3-8B-chat · Standard CIT benchmark | AP: 67.8 -> 69.5 | Improved |
| use OnEDIT instead of GainLoRA | Llama3-8B · Standard CIT benchmark | forgetting rate: 25.2 -> 20.5 | Improved |
| use OnEDIT instead of OLoRA | T5-large · Standard CIT benchmark | AP: 60.1 -> 67.4 | Improved |
| use OnEDIT instead of OLoRA for efficiency | T5-large · adaptation time | time (s): 101.6 -> 42.2 | Improved |
| use OnEDIT instead of LoRA for efficiency | T5-large · adaptation time | time (s): 39.7 -> 42.2 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- InfLoRA — InfLoRA: Interference-Free Low-Rank Adaptation for Continual Learning (CVPR 2024) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- CoIN: A Benchmark of Continual Instruction Tuning for Multimodel Large Language Models (NeurIPS 2024 · similarity 0.88)
- Adapt-$\infty$: Scalable Continual Multimodal Instruction Tuning via Dynamic Data Selection (ICLR 2025 · similarity 0.88)
- Towards Lifelong Model Editing via Simulating Ideal Editor (ICML 2025 · similarity 0.88)
- MAmmoTH2: Scaling Instructions from the Web (NeurIPS 2024 · similarity 0.88)
- MEMOIR: Lifelong Model Editing with Minimal Overwrite and Informed Retention for LLMs (NeurIPS 2025 · similarity 0.88)
- Modality-Decoupled Online Recursive Editing (ICML 2026 · similarity 0.88)
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.
