David is an amazing PhD advisor and everyone should apply.

In collaboration with @CommonCrawl @MLCommons @AiEleuther, the first edition of WMDQS at @COLM_conf starts tomorrow in Room 520A! We have an updated schedule on our website, including a list of all accepted papers.

Super excited for the 1st Workshop on Multilingual Data Quality Signals (WMDQS) which is happening at #COLM2025 tomorrow. We are focused on looking all the way back to the web data that goes into all your LLMs and how we can do better at multilingual. Stop by! @COLM_conf

I finally got around to creating a website for my lab at JHU. Now you can see all the great things my students are doing. https://t.co/Y32jkrabWS

📢 The Joint Call for Tutorial Proposals for EACL/ACL 2026 is now live 👉 https://t.co/VVRLcTbeZb Tutorial proposals should be submitted online via the Softconf system at the following link: https://t.co/156NA7m2bc Deadline: 20 Oct 2025 @eaclmeeting @ #ACL2026NLP #EACL2026

A new study from Johns Hopkins researchers @nikhilsksharma, @ZiangXiao, and @kentonmurray finds that multilingual #AI privileges dominant languages, deepening divides rather than democratizing access to information. Read more:

If you want to help us improve language and cultural coverage, and build an open source LangID system, please register to our shared task! 💬 Registering is easy! All the details are on the shared task webpage: https://t.co/bJs8uY39wo Deadline: July 23, 2025 (AoE) ⏰

📢When LLMs solve tasks with a mid-to-low resource input/target language, their output quality is poor. We know that. But can we pin down what breaks inside the LLM? We introduce the 💥translation barrier hypothesis💥 for failed multilingual generation. https://t.co/VnrOWdNPr8

Fluent, fast, and fair—in collaboration with @MSFTResearch, Johns Hopkins computer scientists (including @fe1ixxu & @kentonmurray) have built a new machine translation model that achieves top-tier performance across 50 diverse languages. Learn more: https://t.co/4vBEEDWIgS

Many LLMs struggle to produce Dialectal Arabic. As practitioners attempt to mitigate this, new evaluation methods are needed. We present AL-QASIDA (Analyzing LLM Quality + Accuracy Systematically In Dialectal Arabic), a comprehensive eval of LLM Dialectal Arabic proficiency (1/7)

If you are at NAACL, check out Tianjian’s presentation of our paper on training with imbalanced data (aka, languages or domains).

Calling for participants in our workshop on retrieving videos. Large pre-trained models do poorly on this and there's a lot of potential to push the field without needing tons of GPUs.

#NAACL2025 30 April - 2pm, Hall 3, Special Theme "Faux Polyglots: A study on Information Disparity in Multilingual Large Language Models". Come visit and learn about how multilingual RALMs fail to handle multilingual information conflicts. Teaser:

Dialects lie on continua of linguistic variation, right? And we can’t collect data for every point on the continuum...🤔 📢 Check out DialUp, a technique to make your MT model robust to the dialect continua of its training HRLs, including unseen dialects. https://t.co/d8DUKcP4gZ

There’s a lot of PhD advisors who are neglecting key parts of their jobs. Your students should NEVER have 100s of words of text on a slide, nor a grainy image of a LaTeX table (bar charts please!) A good talk is now the exception not the norm. It’s all conferences :( #AAAI2025

We now have a Slack channel for the 2025 Workshop on Multimodal Augmented Generation via Multimodal Retrieval! This channel will serve as the primary communication method between authors, participants, and organizers:

@karpathy Perfect timing, we are just about to publish TextArena. A collection of 57 text-based games (30 in the first release) including single-player, two-player and multi-player games. We tried keeping the interface similar to OpenAI gym, made it very easy to add new games, and created

Large Model Inference Efficiency can be tackled from many angles, mixture of experts, efficient self-attention, quantisation, distillation, hardware acceleration.. But what if we could completely avoid redundant computational processing over context window? In our NeurIPS'24

New Workshop on Multimodal Augmented Generation via MultimodAl Retrieval (MAGMaR) to be held at @aclmeeting ACL in Vienna this summer. We have a new shared task that stumps most LLMs - including ones pretrained on our test collection.

📢 Want to host MASC 2025? The 12th Mid-Atlantic Student Colloquium is a one day event bringing together students, faculty and researchers from universities/industry in the Mid-Atlantic. Please submit this very short form if you are interested in hosting! Deadline January 6th