AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

Learning from Itself: Mining Internal Knowledge from Vision Language Models for Continual Learning

Yizheng Gong, Siyue Yu, Waleed Al-Nuaimy, Jimin Xiao

CVPR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Open-vocabulary alignment from VLMs

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 auxiliary CLIP loss with generated tokens to CE (ablation)CLIP finetuning · CIFAR-100 10-tasklast accuracy: 80.91 -> 82.73Improved
add auxiliary CLIP loss with generated tokens to CE (ablation)CLIP finetuning · CUB-200 10-tasklast accuracy: 71.06 -> 74.12Improved
use auxiliary CLIP loss alone without classification loss (ablation)CLIP · CIFAR-100 10-tasklast accuracy: 71.11 -> 75.83No clear effect
add mutual distillation between vision and CLIP heads (ablation)CE + vision CE · CUB-200 10-tasklast accuracy: 71.48 -> 79.98Improved
add vision classifier head without distillation (ablation)CLIP finetuning · CIFAR-100 10-tasklast accuracy: 80.91 -> 81.22No clear effect
use symmetric mutual learning instead of adaptive distillation (ablation)LfI · CIFAR-100 10-tasklast accuracy: 82.73 -> 82.49No clear effect
use adaptive bidirectional distillation over unidirectional (ablation)LfI · CUB-200 10-tasklast accuracy: 80.82 -> 81.74Improved
allow vision head to train during unidirectional distillation (ablation)LfI · CUB-200 10-tasklast accuracy: 76.51 -> 80.82Improved
initialize generated tokens with class name tokens (ablation)LfI token generation · CIFAR-100 10-tasklast accuracy: 82.35 -> 83.82Improved
add COCO reference data in token generation (ablation)LfI token generation · CIFAR-100 10-tasklast accuracy: 82.77 -> 83.82Improved
use generated tokens instead of external MLLM captionsLfI · CIFAR-100 / CUB-200last accuracy: 83.82 vs 84.08; 81.74 vs 82.12No clear effect
use CLIP inference instead of vision-only inference (ablation)LfI · CIFAR-100 10-tasklast accuracy: 82.40 -> 83.82Improved

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