ZurichNLP #8
Brian DuSell from ETH Zurich on stack attention and Leandro von Werra from Hugging Face on BigCode.
Loading…
111 registered
Speakers
Brian DuSell
ETH ZurichLeandro von Werra
Hugging FaceAbout
We are continuing strong in 2024 with Meetup #8 on February 27th! We’re happy to announce the following speakers:
- Brian DuSell, Postdoc @ ETH Zurich
- Leandro von Werra, Chief Loss Officer @ Hugging Face
RSVP soon as spots are limited, the last few times we maxed out capacity. Talk details:
Speaker #1: Brian DuSell Title: Stack Attention: Improving the Ability of Transformers to Model Hierarchical Patterns Summary: Language contains hierarchical syntactic patterns, but transformers do not have a mechanism for dealing with hierarchies of arbitrary depth. In this talk I will present my work on Stack Attention, a kind of self-attention with a latent model of syntax that addresses this limitation and allows transformers to model any context-free language.
Speaker #2: Leandro von Werra
Title: BigCode: Building LLM’s for Code in an Open and Responsible Way
Hope to see you there!
Programme
- Stack Attention: Improving the Ability of Transformers to Model Hierarchical PatternsBrian DuSell
- BigCode: Building LLMs for Code in an Open and Responsible WayLeandro von Werra