Thomas Zeng

[Kangwook Lee @Kangwook_Lee]

🚨New Paper Alert🚨

Could generative agents powered by LLMs transform social science by accurately simulating human social behaviors at scale?

We tested this possibility with virtual humans facing disease threats in "Infected Smallville."

🧵

[Kangwook Lee @Kangwook_Lee]

A new paper on accelerating LLM inference through a novel approach called Draft-based Approximate Inference!
https://t.co/7GDONYkoXb

This framework connects two closely related yet previously distinct areas: approximate inference and speculative decoding.

🧵

[Kangwook Lee @Kangwook_Lee]

🚀 Excited to share our work on Encoder-only Next Token Prediction (ENTP)!

While most successful LLMs are decoder-based, we asked: Can encoder-only TFs be used for next-token prediction?

Yes!

Moreover, ENTP might be better than decoder-only models!!! 😎