Aimed at unlocking the secrets of #AI, Northeastern U. will lead a groundbreaking research project backed by the U.S. National Science Foundation. Read more: bit.ly/4a8isgW

I am delighted to officially announce the National Deep Inference Fabric project, #NDIF. ndif.us NDIF is an U.S. National Science Foundation-supported computational infrastructure project to help YOU advance the science of large-scale AI.

🤔How close are we to fully automating biomedical meta-analyses? Check out our new preprint (w/ @pogrebits3, Iain Marshall, & byron wallace) which investigates how well LLMs can extract numerical results from RCTs to permit accurate, fully automated meta-analysis. 🧵👇

The National Deep Inference Fabric #NDIF, an U.S. National Science Foundation-funded AI research infrastructure project, is awarding 2024 **Summer Engineering Fellowships** in Boston. These are summer visiting positions, for current or recent PhD or undergrads, including stipend, travel and housing costs.

+ Hiba Ahsan!

Excited to present our work on interpretable risk prediction tomorrow @ #NAACL2024! Feel free to stop by our poster from 4-5:30.

Somin has some cool results on CoT + distillation👇

Sheridan has some cool results on tokenization in LLMs and their "implicit vocabularies"👇

There's a general feeling that AI-written text is repetitive. But this repetition goes beyond phrases like "delve into"! We can actually characterize repetition at a structural level using syntactic templates... 🧵 arxiv.org/abs/2407.00211

Finally got to read Chantal ‘s very cool work on syntactic templates in AI generated text. Sharing some findings; The pattern DT NN IN NN doesn’t exist at all in the human written text from the data I am using while it’s extremely common in LLM generations #NLProc

1/10 Excited to share our #BioNLP at #ACL2024NLP paper “Open (Clinical) LLMs are Sensitive to Instruction Phrasings", paper 📄 : arxiv.org/abs/2407.09429…

NNsight and NDIF Democratizing Access to Foundation Model Internals The enormous scale of state-of-the-art foundation models has limited their accessibility to scientists, because customized experiments at large model sizes require costly hardware and complex engineering

Time to study #llama3 405b, but gosh it's big! Please retweet: if you have a great experiment but not enough GPU, here is an opportunity to apply for shared #NDIF research resources. Deadline July 30: ndif.us/405b.html You'll help NDIF test, we'll help you run 405b

I’m very excited to join Northeastern U. Khoury College of Computer Sciences as an assistant professor starting Fall '25!! Looking forward to working with the amazing people there! Until then I'll be a postdoc at NLP @ Uni Vienna with Ben Roth, so reach out if you want to meet up while I'm over in Europe ✨

🚀 New NNsight features launching today! If you’re conducting research on LLM internals, NNsight 0.3 is now available. This update introduces advanced features, offering deeper insights for complex investigations into model behavior. 👇 Here’s what’s new: colab.research.google.com/github/ndif-te…

How can you tell if text is #AI generated? Researchers at Northeastern U. have figured out a new method analyzing sentence structure. Read more: bit.ly/4hthz7o

headed to miami/#EMNLP2024 next week with this work 🌴🌴 come check out the poster (12/11) and let’s talk about templates in text!

[📄] Are LLMs mindless token-shifters, or do they build meaningful representations of language? We study how LLMs copy text in-context, and physically separate out two types of induction heads: token heads, which copy literal tokens, and concept heads, which copy word meanings.