Our results shows how brittle LLMs are at relational reasoning and memory, crucial for knowledge graphs, multi-document analysis, and sense-making.
📢: Accepted in IEEE ICKG'25
https://t.co/rjzbWjrZuv
Shoutout to @ShengzheXu@mandarsharma@profnaren (4/4)
#LLM#NLProc
The Gemini 2.5 technical report is here! It’s been an incredible 9 months since I joined the Gemini team, where our team has been focused on advancing its tool use capabilities. Seeing our models at the top of the leaderboards is a testament to the brilliant team at @Google
Congrats to @SanghaniCtrVT director Naren Ramakrishnan (@profnaren), named @TheOfficialACM fellow (the association's most prestigious member grade) for leadership in developing new algorithms & systems to forecast disease outbreaks & geopolitical events. https://t.co/YceqoXC6mI
Two quick but ✨ major ✨ announcements - I just successfully defended my PhD in LLMs @virginia_tech under @profnaren and will be joining @GoogleAI starting this September! (1/2)
What if adding a few *fixed* lines of code in your LLM training loop could give you a slight performance bump on almost *all* downstream tasks? 👀👀
2min read below (1/4)
What if adding a few *fixed* lines of code in your LLM training loop could give you a slight performance bump on almost *all* downstream tasks? 👀👀
2min read below (1/4)