📢 The #NAACL2022 Call For Papers is out! 2022.naacl.org/calls/papers/ New this year: - Reviewing for main conference submissions will be handled by ACLRollingReview, except for Special Theme submissions. - Optional reproducibility badges!

ACLRollingReview The theme of NAACL 2022 is “Human-Centered NLP”. We invite submissions that address research questions that meaningfully incorporate stakeholders in the design, development, and evaluation of NLP resources, models and systems. More details: 2022.naacl.org/blog/special-t…

BabyBERTa: Learning More Grammar With Small-Scale Child-Directed Language By Philip A. Huebner, Elior Sulem, Cynthia Fisher and Dan Roth

Excited to announce a new product from AWS AI: Amazon CodeWhisperer aws.amazon.com/blogs/aws/now-…

aws.amazon.com/codewhisperer/ is really neat. Helps you code faster, checks for security vulns, discloses licenses of code it drew from, and works great for AWS APIs. Boom! Amazon Web Services putting ML to work for developers

Just out from AWS AI: aws.amazon.com/blogs/machine-…

I’ve been working with Amazon Web Services’s #Bedrock service for a couple of months now at Caylent, and I’d like to share some of what I’ve learned. 🧵

Can GPT-4V and Gemini-Pro perceive the world the way humans do? 🤔 Can they solve the vision tasks that humans can in the blink of an eye? 😉 tldr; NO, they are far worse than us 💁🏻‍♀️ Introducing BLINK👁 zeyofu.github.io/blink/, a novel benchmark that studies visual perception

🚀Introducing "Fewer Truncations Improve Language Modeling" at #ICML2024 We tackle a fundamental issue in LLM pre-training: docs are often broken into pieces. Such truncation hinders model from learning to compose logically coherent and factually grounded content. 👇🧵1/n

The common practice in LLM pre-training is to concat all docs then split into equal-length chunks. This is efficient but hurts data integrity: doc fragmentation leads to loss of info, and causes next-token prediction to be ungrounded, making model prone to hallucination.🧵2/n

Best-fit Packing completely eliminates unnecessary truncations while retaining the same training efficiency as concatenation with <0.01% overhead tested on popular pre-training datasets like Technology Innovation Institute's RefinedWeb and BigCode's Stack.🧵5/n

Can Text-to-Image models understand common sense? 🤔 Can they generate images that fit everyday common sense? 🤔 tldr; NO, they are far less intelligent than us 💁🏻‍♀️ Introducing Commonsense-T2I 💡 zeyofu.github.io/CommonsenseT2I/, a novel evaluation and benchmark designed to measure

🔥Highlights of the Commonsense-T2I benchmark: 📚Pairwise text prompts with minimum token change ⚙️Rigorous automatic evaluation with descriptions for expected outputs ❗️Even DALL-E 3 only achieves below 50% accuracy (2/n)

🔥Error Examples from DALL-E 3 👀More Visualizations: zeyofu.github.io/CommonsenseT2I… (3/n)

😺 This work is done with my amazing collaborators: Yujie Alice Lu, muyu he, William Wang Dan Roth YOU ARE THE BEST!!! 😎🔥 (n/n)

Augmenting GPT-4o with Visual Sketchpad ✏️ We introduce Sketchpad agent, a framework that equips multimodal LLMs with a visual canvas and drawing tools 🎨 . Improving GPT-4o's performance in vision and math tasks 📈 🔗: visualsketchpad.github.io