English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The deployment of Large Multimodal Models (LMMs) within AntGroup hassignificantly advanced multimodal tasks in payment, security, and advertising,notably enhancing advertisement audition tasks in Alipay. However, thedeployment of such sizable models introduces challenges, particularly inincreased latency and carbon emissions, which are antithetical to the ideals ofGreen AI. This paper introduces a novel multi-stage compression strategy forour proprietary LLM, AntGMM. Our methodology pivots on three main aspects:employing small training sample sizes, addressing multi-level redundancythrough multi-stage pruning, and introducing an advanced distillation lossdesign. In our research, we constructed a dataset, the Multimodal AdvertisementAudition Dataset (MAAD), from real-world scenarios within Alipay, and conductedexperiments to validate the reliability of our proposed strategy. Furthermore,the effectiveness of our strategy is evident in its operational success inAlipay's real-world multimodal advertisement audition for three months fromSeptember 2023. Notably, our approach achieved a substantial reduction inlatency, decreasing it from 700ms to 90ms, while maintaining online performancewith only a slight performance decrease. Moreover, our compressed model isestimated to reduce electricity consumption by approximately 75 million kWhannually compared to the direct deployment of AntGMM, demonstrating ourcommitment to green AI initiatives. We will publicly release our code and theMAAD dataset after somereviews\footnote{https://github.com/MorinW/AntGMM$\_$Pruning}.

Large Multimodal Model Compression via Efficient Pruning and Distillation at AntGroup