Navigating the dynamic landscape of algorithmic, efficiency, and safety challenges in Gen-AI personalization is key. Join us as our workshop sparks principled discussions and fosters collaboration among industry leaders, academics, startups, and policymakers. Save the date! 🗓️🚀 https://t.co/AMbU99PMmK

— Vishvak Murahari (@VishvakM) November 17, 2023

QualEval – your LLM data scientist🥼
Quantitative metrics don’t aid in model improvement
Our philosophy – Quality over quantity✅
QualEval qualitatively analyzes tons of data🔍, reasons over mistakes⚠️, and improves your model
Your model dev copilot✈️
Demo:https://t.co/y0EDvjiLV5 pic.twitter.com/26uGYgBD7R

— Vishvak Murahari (@VishvakM) November 7, 2023

I am excited to share that #PrincetonInnovation has featured our innovations around Data Multiplexing, a novel neural network efficiency paradigm, alongside other cutting-edge research from @Princeton this year. pic.twitter.com/ThCcT85CYG

— Vishvak Murahari (@VishvakM) September 28, 2023

We present MUX-PLMs, a family of multiplexed pre-trained language models (PLMs), with a many-fold higher throughput than standard PLMs. Our MUX-BERT models achieve a 2x/5x speedup with a 2-4% drop in absolute performance on GLUE, and a 1-2% https://t.co/JLohqVIHkl… pic.twitter.com/k1X07bpHay

— Vishvak Murahari (@VishvakM) February 27, 2023

New research from @AmeetDeshpande_, @VishvakM and @karthik_r_n featured in @TechCrunch.

Giving ChatGPT a certain "persona" can make it speak like a "bad person," or even Mao Zedong.

The responses, researchers found, can become consistently toxic.https://t.co/vDEmmhVaHu

— Princeton Computer Science (@PrincetonCS) April 17, 2023

Data Multiplexing for Neural Networks🔀
Can neural networks process multiple instances simultaneously as a single mixed input, similar to how radio channels can share bandwidth to carry multiple signals? Surprisingly, we find they can indeed!! https://t.co/OZR1BE1LWQ
📜 [1/6] pic.twitter.com/5uVQ2p7V3T

— Vishvak Murahari (@VishvakM) February 21, 2022