AttentionX (@AttentionX_AI) header
AttentionX (@AttentionX_AI) profile picture

AI Research Lab & Startup Incubator based in Seoul.

Seoul, South Korea
Joined October 2024
Photo by AttentionX (@AttentionX_AI)Photo by AttentionX (@AttentionX_AI)
[1/7] Recruiting for AttentionX Cohort 6 We're an AI research group & startup accelerator based in Seoul from academia and industry. We work on research, open-source projects, and real-world applications to build the next leading AI startups.
1
2
5
887
[6/7] Our research experience: -EMNLP 2024: MERLIN (Multimodal Embedding Refinement) -ACL 2025: sudo rm -rf agentic_security -ICML 2025: Enhanced Language-Image Toxicity Evaluation -ACL 2025: One-Shot is Enough (LLM Multi-Turn Attacks) + many acceptances to WACV, NeurIPS etc.
1
1
489
[7/7] Our startup experience: -Nari Labs (#1 on Hugging Face, #2 on Hacker News) -Typa AI (YC S24) -Hyprnote (YC S25) -AIM Intelligence (backed by Mashup Ventures) -BlueBrown Partners (backed by FuturePlay)
1
276
[1/3] It was a full house yesterday at Krafton HQ for an exclusive AI seminar hosted by @AttentionX_AI with @Krafton_IP !
1
1
3
338
[2/3] Krafton unveiled exciting innovations using sLMs and agent systems for gaming through Smart Zoi and PUBG Ally, while AttentionX’s startups Nari Labs and AIM Intelligence showcased innovations in frontier TTS model training and agent safety.
1
218
[3/3] We’re especially grateful to Director Lee of the Krafton Deep Learning team for taking time out of his busy day to speak with us. Thank you to everyone who could make it!
163
Huge congratulations to AttentionX alum @_doyeob_ for his new TTS model! #1 on HuggingFace and Github 🎉🎉 Go check it out! Looking forward to all the interesting new work he’s cooking with the team 🧑‍🍳
Two undergrads. One still in the military. Zero funding. One ridiculous goal: build a TTS model that rivals NotebookLM Podcast, ElevenLabs Studio, and Sesame CSM. Somehow… we pulled it off. Here’s how 👇
1
194
We are excited to announce to have 2 papers accepted to EMNLP 2024! 1. Towards Efficient Visual-Language Alignment of the Q-Former for Visual Reasoning Tasks 2. MERLIN: Multimodal Embedding Refinement via LLM-based Iterative Navigation for Text-Video Retrieval-Rerank Pipeline
2
144