Presentation of Attention Is All You Need at NIPS 2017
Ashish Vaswani and his team present the groundbreaking 'Attention Is All You Need' paper at the NIPS 2017 conference, introducing the transformer architecture that would revolutionize AI.
Setting
Long Beach Convention Center, a modern conference hall with high ceilings, expansive glass walls, and a large stage area. The room is filled with rows of seating facing a central podium and multiple projection screens displaying neural network diagrams and equations.
Characters
The figures in this scene as an entity network — co-presence links everyone in the moment; speakers who trade lines are bound tighter. Turn the resolution dial to reveal depth the engine actually computed.
TNGF
SELECTED
Ashish Vaswani
primary
A man in his mid-30s with a lean build, short dark hair, and a neatly trimmed beard. He has sharp, intelligent eyes and an expressive face that conveys both intensity and approachability.
Noam Shazeer
primary
A middle-aged man with a lean build, short dark hair, and a neatly trimmed beard. He wears rectangular glasses that give him a studious appearance. His posture is upright, conveying confidence and readiness to engage with the audience.
Niki Parmar
secondary
A young woman in her late 20s, of South Asian descent, with a slender build and shoulder-length black hair neatly tied back. She wears rectangular glasses that give her a studious appearance, and her dark brown eyes are focused intently on her tasks.
Jakob Uszkoreit
secondary
A tall, lean man in his early 30s with short, dark brown hair and sharp, observant eyes. He has a thoughtful demeanor and carries himself with quiet confidence.
Senior Researcher
background
A middle-aged man with a slightly receding hairline, wearing wire-rimmed glasses. His posture is upright but relaxed, showing the confidence of an experienced academic. He has a neatly trimmed beard and wears a thoughtful expression.
Graduate Student
background
A young man in his mid-20s with a lean build, wearing thick-rimmed glasses that slightly magnify his eager eyes. His dark hair is slightly tousled from running his hands through it in excitement.
Dialog
Ashish Vaswani
As you can see here, the transformer architecture completely eliminates the need for recurrence and convolutions—relying solely on attention mechanisms to model relationships in sequential data.
Noam Shazeer
This shift allows for unprecedented parallelization—we're seeing training speeds improve by orders of magnitude compared to traditional RNN approaches.
Ashish Vaswani
The key insight—and this is where we diverged from previous work—was realizing self-attention could directly model all pairwise relationships in a sequence, regardless of distance.
Noam Shazeer
To put it in practical terms: where an LSTM struggles with long-range dependencies, our model maintains perfect information flow across arbitrary distances.
Chat with Characters
Causal neighbors · 103 linked moments
A
2017
· same location
R
2017
· same figure
2017
· same figure
J
1958
· same figure
A
2017
· same figure
N
2017
· same figure
R
2018
· same figure
D
1960
· same figure
R
2018
· same figure
N
2019
· same figure
X
2019
· same figure
P
2018
· same figure
P
2017
· same figure
P
2019
· influences
G
2019
· same era
G
2019
· follows
P
1965
· same figure
R
2019
· same era
R
2019
· follows
P
2019
· same era
P
2019
· follows
P
2017
· same era
P
2017
· follows
P
2017
· same location
P
2017
· same figure
2015
· same era
2
2010
· same era
F
2016
· same era
G
2015
· same era
A
2017
· same era
X
2019
· same era
M
2016
· same era
F
2015
· same era
G
2016
· same era
5
2010
· same era
L
2009
· same era