AI Paper Atlas

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

Extracting Multimodal Learngene in CLIP: Unveiling the Multimodal Generalizable Knowledge

Ruiming Chen, Junming Yang, Shiyu Xia, Xu Yang, Xin Geng

AAAI 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Feature calibration for unseen-class detection · 0 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 the extracted multimodal block to the initialization (ablation)MM-LG · Flickr30k retrieval I2T (12-layer)I2T recall: 23.47 -> 65.18Improved
add the extracted unimodal blocks to the initialization (ablation)MM-LG · Flickr30k retrieval I2T (12-layer)I2T recall: 23.37 -> 65.18Improved
add the extracted multimodal block to the initialization (ablation)MM-LG · CIFAR-100 (12-layer)accuracy: 78.11 -> 87.70Improved
add the extracted unimodal blocks to the initialization (ablation)MM-LG · CIFAR-100 (12-layer)accuracy: 81.15 -> 87.70Improved
add the extracted multimodal block to the initialization (ablation)MM-LG · Flickr30k retrieval I2T (6-layer)I2T recall: 13.12 -> 51.09Improved
replace TLEG linear expansion with the multimodal block learngeneMM-LG vs TLEG · Flickr30k retrieval I2T (12-layer)I2T recall: 61.05 -> 65.18Improved
replace TLEG with MM-LGTLEG · COCO image captioning (6-layer)CIDEr: 39.87 -> 49.80Improved
replace PT-FT with MM-LG initializationPT-FT · COCO image captioning (12-layer)CIDEr: 53.90 -> 59.34Improved
replace PT-FT with MM-LG initializationPT-FT · Flickr30k retrieval I2T (6-layer)I2T recall: 52.76 -> 51.09Hurt
replace PT-FT with MM-LG initializationPT-FT · Flickr30k retrieval I2T (12-layer)I2T recall: 61.53 -> 65.18Improved
use extracted learngene instead of per-scale pre-training (storage)PT-FT · Parameter storage for three scalesParams (M): 151.4 -> 37.4Improved
use one learngene extraction instead of per-scale pre-training (cost)PT-FT · Pre-training costGPU-hours: 1142.4 -> 407.8Improved

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