🧠#LLM agents often fail in ways that are hard to interpret. Our new #VeriLA framework helps humans systematically verify and understand these failures. Catch the presentation by Yoo Yeon Sung@NAACL2025 at the HEAL Workshop at #CHI2025 on April 26th! 📰 megagon.ai/publications/v… #AgenticAI

Attending hashtag#NAACL25? Come find us at Booth #1. Get more info megagon.ai/links/ #LLM #AI #agents #CompoundAI #NLP #HR #HCI #Data #benchmarks #matching #summarization

📣Hayate Iso will present at #NAACL25. Join the discussion at 10:30 a.m. GMT in Hall 3 Poster 4. Read the paper: megagon.ai/publications/e… #fairness #LLM #AI #HR

📢Attending #NAACL? estevamhruschka will present NLP for HR: A Survey today at 10:30 a.m. Find him at Session F: Poster 4.

#NAACL25 come join Pouya Pezeshkpour for his poster presentation on "LLMs Are Not Intelligent Thinkers: Introducing Mathematical Topic Tree Benchmark for Comprehensive Evaluation of LLMs." arxiv.org/abs/2406.05194

How well do language models cope with complexity beyond a single document? As real-world AI applications grow more complex, robust multi-document reasoning becomes critical, and current LLMs are showing signs of strain. In this article, we explore three of our recent research

If you're interested in how LLMs handle the' messy, ambiguous nature of multi-document inputs and where they fall short, this is for you. #NAACL2025 #ICLR2025 #LLMs #MultiDocument #Summarization #Reasoning #NLPResearch #MegagonLabs #AI #MachineLearning #NLP Pouya Pezeshkpour Seiji Maekawa

How can enterprises adopt agentic AI without reinventing their entire tech stack? megagon.ai/blue-designing… In Volume 2 of the Blue article series, we take a deeper look at what it takes to make existing data, models, and services work in agentic workflows and what “agentic for

📢 Find us at #MLSys2025 this week! 📣 Eser Kandogan 🇺🇦 will be presenting at the sponsor talk about #Blue, our open-source agentic compound #AI framework for enterprises 👋 Our team is at our exhibitor booth giving away goodies, ready to talk about job and internship

🚀 𝗥𝗲𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗘𝗿𝗮 In our latest blog post, megagon.ai/blue-designing…, we dive into how Megagon Labs' 𝗕𝗹𝘂𝗲 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 reimagines enterprise architectures to support intelligent,

🔍 𝗖𝗮𝗻 𝘁𝗼𝗱𝗮𝘆’𝘀 𝗟𝗟𝗠𝘀 𝗮𝗰𝗰𝘂𝗿𝗮𝘁𝗲𝗹𝘆 𝗾𝘂𝗲𝗿𝘆 𝗿𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗴𝗿𝗮𝗽𝗵𝘀? 𝗗𝗲𝘀𝗽𝗶𝘁𝗲 𝘁𝗵𝗲𝗶𝗿 𝗴𝗿𝗼𝘄𝗶𝗻𝗴 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀, 𝗹𝗮𝗿𝗴𝗲 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 𝘀𝘁𝗶𝗹𝗹 𝘀𝘁𝗿𝘂𝗴𝗴𝗹𝗲 𝘄𝗶𝘁𝗵 𝗽𝗿𝗲𝗰𝗶𝘀𝗲,

🚨 We are excited to announce that our paper, "FactLens: Benchmarking Fine-Grained Fact Verification", has been accepted at #ACL2025 Findings! megagon.ai/publications/f… 🔍 As #LLMs become increasingly prevalent, ensuring the factual accuracy of their outputs is paramount.