Mars @ipmars

You can retain your normative and well-grounded scientific epistemology and still admit that prompting language models is obviously a form of casting spells.

29 22 345 11K 43

View Details

Mars @ipmars

4 weeks ago

@deanwball Export license for tech has been a thing for certain countries. I wonder if big AI labs already required that.

0 0 0 2K 0

View Details

Mars @ipmars

a month ago

@Techmeme @rafeuddin_ Lmao. Hire a bunch of video gamers, give them a score board.

0 0 12 6K 0

View Details

Mars @ipmars

a month ago

@BjarturTomas Touch grass! Srsly!

0 0 0 24 0

View Details

Mars @ipmars

2 months ago

@bryan_johnson @enhanced_games Where is it,?

0 0 0 14 0

View Details

Mars @ipmars

2 months ago

@SGRodriques is right here. While impatiently waiting for asi to tackle science, I will mark anyone with hyperbolic language as a bot :))

I have spent my entire life working on this and thinking about this for the past 4 years. I don't know what will happen in 20 years, but I can promise you that on the 5-10 year timescale, scientists are not out of their jobs. AI is going to massively accelerate the pace of

41 174 1K 163K 501

0 0 1 25 0

View Details

Mars @ipmars

2 months ago

@m_goes_distance Yep!

0 0 0 954 0

View Details

Mars @ipmars

2 months ago

@sudoingX Did this but need voice interaction with it. Couldn’t figure out a way to Talk to terminus

0 1 1 58 0

View Details

Jessica Nutt @JessicaNutt96

3 months ago

Mythos is not a bad name for a model but it would be better if Anthropic switched to using famous Claudes. Monet, Debussy etc. The final model that achieves AGI would obviously be Van Damme

168 647 9K 384K 356

View Details

Mars @ipmars

3 months ago

This tweet was inspired by me :))

Yaroslav Bulatov @ ICML @yaroslavvb

3 months ago

Google has an internal "let it break" essay about a hero engineer whose hard work ends up being a net negative (by masking the underlying issues). My manager sent me that essay when I was trying too hard to get the collective TensorFlow unit test suite green.

66 439 6K 474K 4K

0 0 0 157 0

View Details

Yaroslav Bulatov @ ICML @yaroslavvb

7 months ago

Gradients are energy-inefficient due to long-range dependencies, yet we lack a viable alternative. What if we crowdsource the discovery of what's next? Starting a weekly in-person reading group in SF. WebGPU, online learning, Hinton. Email [email protected].

9 10 199 38K 137

View Details

Mars @ipmars

5 months ago

I got to see my two yo and two months old nephews for her first time. One’s running full on RL. The other one is sft +rl sometimes hf. The 2 yo is a full human with an exceptionally good command of language. Looking for some cognitive experiments to run with him :))

0 0 2 170 0

View Details

Jason Sheltzer @JSheltzer

5 months ago

AI is cool and all... but a new paper in @ScienceMagazine kind of figured out the origin of life? The paper reports the discovery of a simple 45-nucleotide RNA molecule that can perfectly copy itself.

177 1K 6K 868K 3K

View Details

Mars @ipmars

6 months ago

Most can be pruned with one or two questions.

0 0 0 56 0

View Details

Mars @ipmars

6 months ago

First month in the bay. Lots of buzzwords flying around. Not much substance. People are either new ish to ML and use buzzwords or old timey and like simple concepts.