Colin Cherry

Externally retrieving knowledge empowers LLMs for domain-adapted MT ⚖️🩺. But how is knowledge best represented, and how viable is generating it from an LLM itself? Our @GoogleAI paper investigates these questions through a careful experimental setup 📜. arxiv.org/abs/2503.05010

<<Call for BoF/Affinity Group meeting>> Applicants should fill out the application form before March 24 2025.naacl.org/calls/affinity/ #NAACL2025

😼SMOL DATA ALERT! 😼Anouncing SMOL, a professionally-translated dataset for 115 very low-resource languages! Paper: arxiv.org/pdf/2502.12301 Huggingface: huggingface.co/datasets/googl…

The call for Diversity and Inclusion Subsidies is out: #NAACL2025" target="_blank" rel="nofollow noopener">2025.naacl.org/calls/dei_subs…

LLMs are typically evaluated w/ automatic metrics on standard test sets, but metrics + test sets are developed independently. This raises a crucial question: Can we design automatic metrics specifically to excel on the test sets we prioritize? Answer: Yes! arxiv.org/abs/2411.15387

New application link! google.com/about/careers/… I am at EMNLP/WMT this week. Please come find me if you want to learn more about this role!

📢Don't miss the NAACL Student Research Workshop! 🖇️ CFP & Important dates: naacl2025-srw.github.io/cfp #NLProc

Thank you to those who participated in our recent all-member vote regarding our name change. The change is happening! We are: The Nations of the Americas Chapter of the Association for Computational Linguistics! Announcement 👉 naacl.org/posts/2024-10-…

📢 NAACL needs Reviewers & Area Chairs! 📝 If you haven't received an invite for ARR Oct 2024 & want to contribute, sign up by Oct 22nd! ➡️AC form: forms.office.com/r/8j6jXLfASt ➡️Reviewer form: forms.office.com/r/cjPNtL9gPE Please RT 🔁 and help spread the word! 🗣️ #NLProc @ReviewAcl

Interested in doing research on Google Translate and Gemini? Good news! I’m hiring for full-time roles on the Google Translate Research Team! Apply here: google.com/about/careers/…

Researchers from @Google reveal that verbose #LLMs, 🤖 which offer multiple translations 🔄 or refuse to translate, 🚫 pose significant challenges ⚠️ to traditional #MT evaluation frameworks. #machinetranslation @ebriakou @ColinCherry @markuseful slator.com/google-finds-r…

📢 Call for demos is out!! #NAACL2025 #NLProc Check the website for submission guidelines and a chance to win the Best Demo Award! 🏆 🖇️ 2025.naacl.org/calls/demo/

📢📢🌟@jhuclsp Have an Idea? Let’s Hear It! JSALT 2025 Call for proposal is out. Deadline: October 15th, 2024 For more information: clsp.jhu.edu/the-11th-frede…

[1/5] Are verbose #LLM translations skewing evaluation results? TLDR: Yes! Our recent work dives into the prevalence and impact of LLM verbosity in automatic and human evaluations. 📎 Paper: arxiv.org/pdf/2410.00863

📢 Second call for papers is out!! #NAACL2025 #NLProc 🖇️ 2025.naacl.org/calls/papers/

Translation is a complex task involving pre-translation research and post-translation stages. Can #LLMs handle this process step-by-step, relying solely on their internal knowledge? ✨We show that decomposing the translation process significantly improves #Gemini translation quality of long-form texts across all #WMT24 languages! 📜arxiv.org/pdf/2409.06790

📢 Calling all #NLProc enthusiasts! Submit your tutorial and workshop proposals to 2025 *ACL conferences (NAACL, ACL, EMNLP) through one joint call! Tutorials: 2025.naacl.org/calls/tutorial… Workshops:2025.naacl.org/calls/workshop…

🥳 LLMs are changing the game, even for datasets! NewsPaLM, a publicly released LLM-generated dataset, outperforms larger web-crawled corpora for MT. It includes sentence & paragraph-level, MBR-decoded data. See paper for more, incl. LLM self-distillation. arxiv.org/abs/2408.06537

First call for papers is out! #NAACL2025 🔴2025.naacl.org/calls/papers/

[New paper] If you are sampling multiple outputs from a teacher LLM (e.g., Gemini 1.5 GPT), ranking them, and fine-tuning the student on the best output, you can do better. Simple idea: Fine-tune / Distill on the top-k outputs instead. Consistent gains on machine translation.