señor swesearcher @ Google.musician. from 🇪🇨 living in 🇨🇦. mostly share stuff i make, not opinions i have.
MUSICODE: https://t.co/BronryzR7E
website: https://t.co/laUfOvyVDT
Feb 23, 2022 • 8 tweets • 2 min read
it's perf season at google, which is often a big source of stress for lots of us, especially when going for promo.
a few months ago i had an epiphany (of sorts) that got rid of pretty much all my perf anxiety.
a thread in case it helps others (google or not):
1/🧵
a few months ago i was brainstorming with my manager on how i could get to the next level in my ladder.
we came up with lots of ideas, but they all required changing my research direction and/or starting to manage people.
i'd be fine w/ either, but i felt weird about it...
2/🧵
Dec 8, 2021 • 10 tweets • 5 min read
Last year we showed that deep RL can successfully control stratospheric balloons in the real world go.nature.com/3dtvrPr
Today we're announcing a β-release of the BLE, a high-fidelity simulator of this complex real-world decision making problem. bit.ly/3y1i2aK
1/🎈
This sim is based on the one we used in our Nature paper published last year: go.nature.com/3dtvrPr
A faithful simulator was key to our success and proved to be a generally interesting RL problem.
Over the past year we have worked hard to make it accessible to everyone.
2/🎈
Happy to announce that our #RL library, dopamine, now also has JAX implementations of all our agents, including a new agent: QR-DQN! github.com/google/dopamine
1/X
The JAX philosophy goes very well with that of Dopamine: flexibility for research without sacrificing simplicity.
I've been using it for a while now and I've found its modularity quite appealing in terms of simplifying some of the more difficult aspects of the agents.
2/X