AI Paper Atlas

Research map / Continual learning / Continual fine-tuning of LLMs

Task-Distributionally Robust Data-Free Meta-Learning

Zixuan Hu, Yongxian Wei, Li Shen, Zhenyi Wang, Baoyuan Wu, Chun Yuan, Dacheng Tao

TPAMI 2025 · Continual learning › Continual fine-tuning of LLMs · approach: Synthetic data and merging for CLIP · 1 citations (OpenAlex)

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 changedWhereMeasuredEffect
add memory buffer to vanilla synthetic-task training (ablation)TDFML · CIFAR-FS 5-way 1-shotLAST accuracy: 26.76 -> 28.33No clear effect
add task interpolation to memory buffer (ablation)TDFML · CIFAR-FS 5-way 1-shotLAST accuracy: 28.33 -> 38.42Improved
add soft-label supervision from pre-trained models (ablation)TDFML · CIFAR-FS 5-way 1-shotBEST accuracy: 38.76 -> 40.80Improved
add automatic model selection (deceptive-label models, 40% pollution) (ablation)TDFML · CIFAR-FS 5-way 5-shotaccuracy: 47.03 -> 49.72Improved
add automatic model selection (deceptive-accuracy models, 80% pollution) (ablation)TDFML · CIFAR-FS 5-way 5-shotaccuracy: 42.95 -> 45.85Improved
use task mixup hybrid strategy (+mixup) (ablation)TDFML-ProtoNet · CIFAR-FS 5-way 1-shotBEST accuracy: 40.80 -> 41.91Improved
use task mixup hybrid strategy (+mixup) (ablation)TDFML-ProtoNet · CUB 5-way 5-shotLAST accuracy: 47.64 -> 48.78Improved
use ProtoNet instead of ANIL as meta-learner (ablation)TDFML · CIFAR-FS 1-shot vs MiniImageNet 5-shotBEST accuracy: CIFAR-FS 1-shot 40.39 -> 40.80; MiniImageNet 5-shot 43.63 -> 42.93Mixed
replace PURER-ProtoNet with TDFML-ProtoNetmeta-learner · CIFAR-FS 5-way 1-shotLAST accuracy: 27.01 -> 40.28Improved
increase pollution rate from 10% to 80% (deceptive label)PURER · CIFAR-FS 5-way 5-shotaccuracy: 48.81 -> 41.02Hurt
compare with data-free KD method MADTDFML · CIFAR-FS 5-way 5-shotBEST accuracy: 35.19 -> 58.25Improved
compare with continual meta-learning SDMLTDFML · CIFAR-FS 5-way 5-shotBEST accuracy: 49.25 -> 58.25Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy