Your Guide to Abaka AI at EMNLP 2025
Abaka AI team is thrilled to be in Suzhou for EMNLP 2025! As a proud sponsor of the conference, we are excited to connect with the world's leading minds in Natural Language Processing and discuss the critical role of high-quality data in the new age of Agentic AI.
From our booth on the expo floor to our exclusive afterparty and contributed research, hereβs how you can connect with our team in Suzhou.
Meet Abaka AI's Data Experts at Booth #C3-13
Come and meet our team at the Suzhou International Expo Center. We will be at Booth #C3-13 for the main exhibition, ready to discuss the critical data challenges facing today's LLMs.
We believe that Better Data = Better Models. Stop by to explore how our high-quality data, annotation, and model evaluation services can power your next breakthrough in NLP.
EMNLP After Dark: Our Exclusive Happy Hour
Beyond the conference floor, we invite you to continue the conversation in a more relaxed setting. Abaka AI is proud to co-host "EMNLP After Dark" with our partners at SGLang (SGL) and 2077AI.
This is an exclusive gathering for an evening of honest, practical discussion. We're bringing together top MLEs, Tech Leads who have brought real RAG systems to production, and NLP veterans who can help you avoid costly detours. Join us for authentic German food, craft beer, and stunning views of Jinji Lake, just 2 minutes from the venue.
π
Date & Time: Wednesday, November 5th, 6:00 PM - 11:00 PM.
π Location: A unique venue 2-min drive from EMNLP (Address provided upon registration approval)
*This is a closed-door event, and registration is subject to approval. Spaces are limited.
πRequest Your Invitation Here
Discover Our Contributed Research
As part of our commitment to advancing open-source innovation, our contributed research organization, 2077 AI, is presenting a new poster on AgentThink.
This pioneering framework addresses critical challenges for autonomous driving models by integrating Chain-of-Thought (CoT) reasoning with dynamic, agent-style tool invocation. The AgentThink framework is proven to significantly boost overall reasoning scores by 53.91% and enhance answer accuracy by 33.54% on the DriveLMM-o1 benchmark. We invite you to explore this work, which highlights a promising path for developing more trustworthy and tool-aware autonomous driving models.