Ian Thompson

82 posts

Ian Thompson banner
Ian Thompson

Ian Thompson

@iantbd

ai scaling + simplifying. founder @LatentSpaceAI spoilers from my favorite story: “Overhead, without any fuss, the stars were going out”

San Francisco, CA Katılım Şubat 2014
1.5K Takip Edilen421 Takipçiler
Ian Thompson retweetledi
Kanjun 🐙
Kanjun 🐙@kanjun·
Avalon is the world's fastest 3D simulator for RL agents. All baselines train on 1 GPU in ~1 day. We want academic researchers to be able to study aspects of intelligence missing from today’s models, even w/o access to large-scale compute. Get started: generallyintelligent.com/avalon/
GIF
English
2
38
185
0
Ian Thompson retweetledi
Kanjun 🐙
Kanjun 🐙@kanjun·
In 2020, @michael_nielsen & I began a 2-month project to write: "how would we fund science?" 2 years & 40,000 words later, it's become: "how can the culture & institutions of science actually change, and ultimately become self-improving?" Our answer: scienceplusplus.org/metascience/in…
English
28
171
849
0
Ian Thompson retweetledi
Bluesky
Bluesky@bluesky·
Bluesky is building a social protocol. We released “ADX” (the X stood for Experiment) in May. Now that the design is starting to solidify, we’re renaming it to the “Authenticated Transport Protocol” — the “AT Protocol.” blueskyweb.xyz/blog/10-18-202…
English
537
2K
6K
0
Lukas Biewald
Lukas Biewald@l2k·
@zacharylipton Don't know about paper reviews but we find shap pretty useful in practice. I think I'm missing some context - why do you find these tools so meaningless?
English
2
0
8
0
Zachary Lipton
Zachary Lipton@zacharylipton·
Insisting that authors include a LIME/SHAP/IG/TCAV/GRADCAM saliency map in a paper shd be a disqualifying offense for reviewers. Including any such map without a powerful disclaimer (“this means nothing”) shd be a disqualifying offense for authors Extra true for ML in healthcare
English
28
52
405
0
Ian Thompson retweetledi
Ethan Perez
Ethan Perez@EthanJPerez·
It takes a lot of human ratings to align language models with human preferences. We found a way to learn from language feedback (instead of ratings), since language conveys more info about human preferences. Our algo learns w just 100 samples of feedback. Check out our new paper!
Jérémy Scheurer@jeremy_scheurer

Can we train LMs with *language* feedback? We found an algo for just that. We finetune GPT3 to ~human-level summarization w/ only 100 samples of feedback w/ @jaa_campos @junshernchan @_angie_chen @kchonyc @EthanJPerez Paper: bit.ly/377VFaB Talk: bit.ly/3vVJy8V

English
2
25
148
0
Ian Thompson
Ian Thompson@iantbd·
@keerthanpg @jxmnop Half of that in a year would be only 2-3% of Nvidia’s datacenter revenue. But I agree it may have a big effect for Nvidia! It speeds up the compute arms race by creating fomo for companies (like Meta) to spend even more than that to catch up… begun, the compute wars have
English
2
0
1
0
Ian Thompson retweetledi
Jason Yosinski
Jason Yosinski@jasonyo·
Excited to share a peek at what I've been working on recently: we're building networks of little sensors and using ML to help wind farms produce more energy!
Windscape AI@Windscape_AI

Hello World! We are Windscape AI, and we're working to help wind farms produce more clean energy for the planet (and more revenue for their owners). More wind energy == a faster transition away from GHG-emitting fossil fuel energy sources! Quick 🧵 on how it works:

English
6
16
164
0
Ian Thompson retweetledi
Jay Graber 🦋
Jay Graber 🦋@arcalinea·
It’s technically possible for our online identities and relationships to belong to us the way an email address, phone contact book, or password keychain does. So I think it will happen, one way or another. It’s just a matter of time.
English
15
19
165
0
Ian Thompson
Ian Thompson@iantbd·
@iamtrask In the case of AI, this means focusing on key safety problems like alignment (and privacy!) as we scale
English
0
0
0
0
Ian Thompson
Ian Thompson@iantbd·
@iamtrask That’s true, we don’t have the counterfactual :) I think the best outcome would have been to push forward with a promising technology, but with a heavy focus on safety (as opposed to halting development). As far as I know, France did a good job on this front with nuclear energy.
English
1
0
0
0
Thu Nguyen-Phuoc
Thu Nguyen-Phuoc@thunguyenphuoc·
I successfully defended my PhD thesis today! I can't believe it actually happened! Words cannot describe how happy and emotional I'm feeling right now! From an architecture student to a research scientist in computer science. That was wild!!!
Thu Nguyen-Phuoc tweet media
English
38
3
604
0
Leo Gao
Leo Gao@nabla_theta·
Leo Gao tweet media
ZXX
32
94
563
0
Ian Thompson
Ian Thompson@iantbd·
...though Newton did spend ~20% of his time working on alchemy, and we've had some open problems in AI alchemy for a while 😂 youtube.com/watch?v=x7psGH… (4/4)
YouTube video
YouTube
English
0
0
3
0
Ian Thompson
Ian Thompson@iantbd·
Would Darwin/Newton gravitate more towards AI or crypto today? Newton seems like he’d be more interested in snark hunting, while Darwin would work on scaling laws. (3/4)
English
1
0
2
0
Ian Thompson
Ian Thompson@iantbd·
Here’s another memetic isomorphism: Crypto is more like physics, and AI is more like biology – not in “biologically inspired” sense, but in the nature of possible solutions. So using the wordcel / rotator basis, Darwin is a wordcel and Newton is the ultimate rotator. (1/4)
vitalik.eth@VitalikButerin

16. It's fascinating how the two axes seem to imply each other: Pure tinkering is fast and effective, but it is unprincipled, so lacks large-scale legitimacy, so can only cooperate through centralization Theorycel thinking is slower, but well-suited for decentralization

English
1
1
5
0

Keşfet

@michael_nielsen @l2k @zacharylipton @keerthanpg @jxmnop @iamtrask @thunguyenphuoc @elonmusk