Fernanda Viégas

Human/AI Interaction and data visualization. Professor at Harvard. Co-founder of Google's People+AI Research initiative (PAIR)

[Fernanda Viégas @viegasf]

Great work by @ghandeharioun and others in PAIR on a new interpretability framework: 🩺Patchscopes

[Asma Ghandeharioun @ghandeharioun]

🧵Can we “ask” an LLM to “translate” its own hidden representations into natural language? We propose 🩺Patchscopes, a new framework for decoding specific information from a representation by “patching” it into a separate inference pass, independently of its original context. 1/9

[Fernanda Viégas @viegasf]

Here's a close-up of hue and brightness heads in a vision transformer! Work done with tremendous help from Aoyu Wu and @chenxcynthia

[Fernanda Viégas @viegasf]

Visualize transformer attention! ✨

AttentionViz, created by Catherine Yeh and expanded by Yida Chen, helps you explore transformer self-attention by visualizing query and key vectors in a joint embedding.

Paper: https://t.co/M7upXjDoZG
Website: https://t.co/rKJgIHm6q0

[Fernanda Viégas @viegasf]

Language models have some beautiful spiral plots reflecting positional patterns. And a vision model has heads that arrange images according to brightness and hues. But there’s a lot more to find! What else can you see?

[Fernanda Viégas @viegasf]

Toasters have blinking lights, cars have speedometers. Should chatbots have dashboards too?

A speculative essay👉 The System Model and the User Model: Exploring AI Dashboard Design
https://t.co/HLcjX3Kdwe

[Harvard SEAS @hseas]

Students explore the aesthetics of computing in new computer science course at SEAS. "CS73: Code, Data, and Art" is co-taught by @wattenberg and @viegasf, and teaches students how to create abstract art and communicate data sets through visualizations. https://t.co/se8D0Tkbqd

[Fernanda Viégas @viegasf]

The Wind Map turns 10 today! I'm still mesmerized by the powerful force it depicts. https://t.co/Ay8hhTNWsK

[Rachel Bara @rachsturges]

Ten years ago #OTD, @viegasf and @wattenberg released "wind map." Then came @cambecc's https://t.co/B53U1wU2pU.

My story for Earth Island Journal.

https://t.co/AbarRZ7buN

[Fernanda Viégas @viegasf]

Looking for a postdoc position in data visualization, human-centered machine learning, or both? Consider joining our lab at Harvard!

Here’s the official announcement: https://t.co/ztF5ByJkzW

[Fernanda Viégas @viegasf]

My kids just donated to @SavetheChildren to support their life changing programs for children in need https://t.co/GKbdlLbTu5

Consider donating this season.

[Fernanda Viégas @viegasf]

And for information on the application process, see: https://t.co/8ZTXfGc2UX

[Fernanda Viégas @viegasf]

Thinking about grad school next year? Interested in visualization, human/AI interaction, computational art?

Consider Harvard! @wattenberg and I are building up our lab and will be looking for students.
Our DMs are open.

[Fernanda Viégas @viegasf]

@whaleandpetunia @wattenberg Yes, we are looking for PhD candidates..

[Fernanda Viégas @viegasf]

@omarckal @wattenberg I believe it should be working now

[Meredith Ringel Morris @merrierm]

We are hiring a full-stack software engineer for the PAIR team at @GoogleAI to work on projects at the intersection of #ML and #HCI, including our interpretability tools to support responsible #AI practices. Apply at https://t.co/bk80pUR2yw

[Fernanda Viégas @viegasf]

3/3
"In Texas, they like to buy land, beer" vs
"In New York, they like to buy art, books"

https://t.co/tO2pJQJV7v

[Fernanda Viégas @viegasf]

1/3
"We went to Brazil because we wanted to escape" vs
"We went to Japan because we wanted to study"

Check out @adamrpearce's language model explorable, showing what models like BERT have learned about the world. #googlePAIR

Type your own queries:
https://t.co/tO2pJQJV7v

[Fernanda Viégas @viegasf]

2/3
"Fernanda was born in the year of 1954" vs
"Martin was born in the year of 1750"

At least it thinks I'm younger than @wattenberg... 😂

[Fernanda Viégas @viegasf]

Hannah Fry @FryRsquared does a great job of showing how "humble" graphs such as line charts and scatter plots have changed the world and our understanding of it:

Thanks for the pointer, @m_pellat !

https://t.co/HgNUnBFf62

[Fernanda Viégas @viegasf]

Are you an academic working in the areas of Participatory ML, Accessibility, Algorithmic Fairness or Higher Education in CS? If so, apply to Google’s Award for Inclusion Research (AIR)!

AIR is an international program 🤩
Application deadline: July 14
https://t.co/HGz4Czbggh

[Nikhil Thorat @nsthorat]

Interested in working with the PAIR team at Google on projects like Know Your Data?

Apply here, we're hiring!

And yes, "full stack": we really care about folks thinking about the intersection between UX / frontend and large scale data munging.

https://t.co/emMOlHSoF7