![Genta Winata Profile](https://pbs.twimg.com/profile_images/1437097742208208898/p3uoRkVv_x96.jpg)
Genta Winata
@gentaiscool
Followers
2K
Following
7K
Statuses
1K
AI Researcher @CapitalOne AIF. Ex @TechAtBloomberg @BigScienceW @SFResearch @hkust. Working on multilingual and LLM #NLProc. Building @GrassrootsSci
Joined December 2011
π World Cuisines πππππππ’π‘ π Big news! Thrilled to share that our WorldCuisines paper has been accepted by @naaclmeeting NAACL 2025! βοΈ This is the first-ever large-scale high-quality benchmark for multilingual VQA on Food, covering 30 languages and dialects with over 1 million data points. Check our paper, dataset, website, and leaderboard! π€ π° π π
Thank you for all collaborators @fredyhudi, @patrickamadeus_, @davidanugraha, @rifkiaputri, @zzeet, @ubaidalih, @auliaadilaa, @adamnohejl, @JunhoMyung00211, @aliceoh, @AnarSnowball, @faridlazuarda, @jcblaisecruz, @nedjmaou, @jodieyzhou, @AboladeDaud, @prajdabre1, @holylovenia, @SCahyawijaya, @bryanwilie92, @mrpeerat, @farizikhwantri, @gkuwanto, @llamagrp, @mv_zhukova, @EmmanueleChers1 @AlhamFikri @davlanade @tarowatanabe, and to the many others not on X, whose tireless contributions and language expertise were crucial to the success of this project. #worldcuisines #nlproc #NAACL2025
#food #vlm #llava #nlvm #Pangea
π World Cuisines π£ππ₯§π¨π±ππͺπ₯π« π€ Is this dumpling π₯ known as gyoza, jiaozi, or momo? is it a Chinese, Japanese or Korean dish? We are excited to introduce WorldCuisines, a high-quality 1β£ million massive-scale multilingual parallel VQA benchmark available in 30 languages, dialects, and variations. This benchmark is designed to assess the VLM's capability to comprehend cultural nuances in global cuisines. Check this thread π§΅π π€ π° π π
#worldcuisines #nlproc @aclmeeting #food
#vlm #llava #aira #nlvm (1/n)
2
15
57
RT @WCALCS: π£ The paper submission deadline has been extended to February 10, giving you additional time to finalize and submit your paper.β¦
0
1
0
RT @dbaek__: 1/9 π¨ New Paper Alert: Cross-Entropy Loss is NOT What You Need! π¨ We introduce harmonic loss as alternative to the standard CEβ¦
0
548
0
RT @devarda_a: New preprint! π§ π€ Brain encoding in 21 languages! w/ @saima_mm, @GretaTuckute, and @ev_fedorenko (β¦
0
31
0
RT @WCALCS: π£ Paper Submission Deadline β Feb 7! Don't miss out! make sure to submit your paper before the deadline for CALCS @naaclmeetinβ¦
0
2
0
π RainbowPO βοΈ I'm excited to congratulate my intern at @CapitalOne, @OptionsGod_lgd, whose paper was accepted by @iclr_conf #ICLR2025! Itβs been a pleasure mentoring you. A huge shout-out to @zzeet, Shi-Xiong Zhang, David Yao, Wenpin Tang, and Sambit Sahu. Check our preprint: Openreview: #NLProc #DPO #RainbowPO
π RainbowPO βοΈ I'm excited to share the work of my intern, @OptionsGod_lgd, at @CapitalOne on Human Preference Alignment. We introduce π RainbowPO, a unified offline preference optimization framework that combines and enhances the best aspects of all DPO methods. It outperforms SimPO, ORPO, CPO, KTO, IPO, and vanilla DPO. Check the paper π° For more details, check out @OptionsGod_lgd's thread 𧡠#NLProc #Alignment #DPO #RainbowPO #simpo #xpo #kto #ipo #orpo #cpo #humanpreference #rlhf
1
4
23
@naaclmeeting We would also like to express our gratitude to @yueqi_song and @xiangyue96 for their assistance with replicating the Pangea model, as well as to @Wenliang_Dai and @nayeon7lee for their contributions to the NVIDIA NVLM model.
0
0
7