Yue Zhou

PhD Student in NLP@UIC. #NLProc for Social Good, Healthcare #NLP #DeepLearning Amateur Foil Fencing Player. Classical Guitar. Snowboarding.

[Yue Zhou @DataFleuret]

Excited to share two papers exploring and exploiting LLMs deficiencies accepted to #EMNLP2024 and #NAACL2024! 🎉
1. LLMs Are Involuntary Truth-Tellers: https://t.co/hNS6NtA9eT
2. Surface Form Impact on Math Reasoning:
https://t.co/kgR9ICuL5L
#LLMs #NLProc

[Michael Black @Michael_J_Black]

In the LLM-science discussion, I see a common misconception that science is a thing you do and that writing about it is separate and can be automated. I’ve written over 300 scientific papers and can assure you that science writing can’t be separated from science doing. Why? 1/18

[UIC CS Department @UICCS]

Congratulations to Barbara Di Eugenio for receiving the @AWISNational Zenith Award for her lifetime achievements in STEM and her commitment to workplace diversity. https://t.co/p95pR54pkU

[Tal Linzen @tallinzen]

Just to make sure this point doesn't get lost: the success of instruction-finetuning highlights the *limitation* of the self-supervised language modeling objective.

[Yue Zhou @DataFleuret]

This is.. amazing!

[Christian Bluethgen @cxbln]

🎉 #StableDiffusion can be fine-tuned to generate medical images, and the outputs can be controlled using natural language text prompts!

In our latest work, we use SD to create synthetic chest xrays and insert pathologies like pleural effusions.

🧵 #Radiology #AI #StanfordAIMI

[Stable Diffusion @StableDiffusion]

Over the 7 weeks since Stable Diffusion's release, we've seen many amazing open-source contributions from the community. A lot of them have come in the form of awesome Google Colab notebooks! 🔥

Here is a thread of 14 awesome notebooks we've seen from the community ↓

[Nanyun (Violet) Peng @VioletNPeng]

Excited to share our #Neurips2022 paper on controllable text generation with theoretical guarantees. We decompose a sequence-level oracle into token-level guidance to steer the generation to consider future constraints. Impressive results for incorporating OOVs.

[Hanna Hajishirzi @HannaHajishirzi]

My alma mater, Sharif University of Technology, Iran's premier university, was under siege yesterday. Many students were arrested, heavily injured or perhaps killed. Many Iranian scholars and PhD students whom you know have received their BSc degrees from this university.

[Christian Hubicki @chubicki]

Elon Musk revealed the #TeslaBot today and I have #robotics thoughts.

https://t.co/BMyNcjI12J
🧵

[Graham Neubig @gneubig]

I was a bit short on research ideas, so I decided to ask @chrmanning (as simulated by @huggingface 's BLOOM https://t.co/JNldgY4Qwn) for some inspiration. The advice was...

[Yue Zhou @DataFleuret]

@chrmanning Even though the generated text doesn’t make much sense…I can still imagine your voice when I am reading it 😂 especially from “of course you can! This is like asking..” would be even better if it ended with “whatsoever” 😀

[Samuel "curry-howard fanboi" Ainsworth @SamuelAinsworth]

📜🚨📜🚨
NN loss landscapes are full of permutation symmetries, ie. swap any 2 units in a hidden layer. What does this mean for SGD? Is this practically useful?

For the past 5 yrs these Qs have fascinated me. Today, I am ready to announce "Git Re-Basin"!

https://t.co/mRu5k3ONUm

[Chris Olah @ch402]

I've never had so many "this can't possibly be true, we must have a bug" results in the course of a research project before.

I'd like to take a moment to walk through some of the very strange (and surprisingly beautiful) things we found.

[Yann LeCun @ylecun]

PyTorch is now handled by an independent foundation, the #PyTorchFoundation, which is under the #LinuxFoundation umbrella.

[David McAllester @McAllesterDavid]

I have long argued that grounding is not necessary for understanding. I laid out my case against grounding in a response to Browning and LeCun. @ylecun @ilyasut @jurafsky @manning @percyliang
https://t.co/VlKA5rYROH

[Graham Neubig @gneubig]

We've started the Fall 2022 edition of:
🎓CMU CS11-711 Advanced NLP!🎓

Follow along for
* An intro of core topics
* Timely content; prompting, retrieval, bias/fairness
* Content on NLP research methodology

Page: https://t.co/wxXDhSiVqf
Videos: https://t.co/XQPzc6ss99

[Sam Bowman @sleepinyourhat]

I've had a similar experience to Ethan. If you want to do an NLP data collection/labeling process and don't want/need to be managing annotators directly, Surge is remarkably easy to work with and their team does very good work.

[David McClure @clured]

Inspired by various recent efforts to make sense of the text2img datasets - here's all 12M captions from LAION-Aesthetics with score > 6, embedded with CLIP and UMAP'ed to 2d. Color is the domain of the image URL.

[Richard Socher @RichardSocher]

Nobody is working on this seriously. Large language models have no goals, they just learn to predict the next item in a sequence.
We, as a species and research community, are surprisingly uncreative when it comes to identifying useful objective function spaces.

[Mark Tenenholtz @marktenenholtz]

DALLE-2 was paywall-released recently by an extremely well-funded company.

Just yesterday, a group of independent researchers released their own model (Stable Diffusion) that you can use in a few lines of code for FREE.

The speed of the ML research community is insane 🤯