Lightnews — Scholar-powered news

Jonathan Balloch @balloch.bsky.social · Sep 8

Ooo peak design is legit

Jonathan Balloch @balloch.bsky.social · Mar 25

for the record, this is why LLMs have been more widely successful and applicable than, say, vision-language-action models, and why VLAs are catching up: this is a recipe that can be applied very broadly, but only works at a production level if the data domain is VERY thoroughly covered

Jonathan Balloch @balloch.bsky.social · Mar 25

The more data you have, the better an embedding space you have, and the more likely your interpolation is to be correct. So you are right in that the something like the answer is probably in the training data, but you are wrong that the exact answer is in the training data or searched for.

1 2

Jonathan Balloch @balloch.bsky.social · Mar 25

Like many social media discussions, what is missing here is nuance. LLMs, like all generative no-prior ML models, are, effectively, interpolating. But in the case of LLMs, they are interpolating in the space of "next token embedding."

1

Jonathan Balloch @balloch.bsky.social · Mar 25

Fundamentally you can *have* both, but functionally when you optimize for multiple objectives usually only one ends up as the primary. Guzdials article is suggesting that the prior push being so attached to undergrad outcomes is a bad primary objective for K-12 students, which is reasonable...

1

Jonathan Balloch @balloch.bsky.social · Mar 25

Le Chat underrated

Remi Cadene @remicadene.bsky.social · Mar 25

I absolutely love @MistralAI. Even the free version is super fast and report sources 🤩 The pro version is even better.

Jonathan Balloch @balloch.bsky.social · Mar 25

well thats not great

carpenter22.bsky.social @carpenter22.bsky.social · Mar 25

Too busy with costumes. As the Carolinas burn, Homeland Security head Kristie Noem says Trump administration is going to ‘eliminate’ FEMA www.the-independent.com/news/world/a...

Kristie Noem says Trump administration is going to ‘eliminate’ FEMA

Poor, disaster-prone red states would likely be hit hardest if the federal government withdraws funding

www.the-independent.com

Reposted by Jonathan Balloch

Cheng Soon Ong @ml4x.bsky.social · Mar 22

I think a deeper difficulty in ML is the economy of attention. The hundreds of papers each day released on ArXiv in ML means that a reader needs to resort to heuristics to keep up. Stuff like trust a recommender system, or only read famous authors, or scan for buzzwords.

2 1 5

Jonathan Balloch @balloch.bsky.social · Mar 17

Sarah Paine is incredible

Jonathan Balloch @balloch.bsky.social · Mar 11

Given whats going on in the world, I think its time to reread Brave New World

1

Reposted by Jonathan Balloch

Christian Wolf @chriswolfvision.bsky.social · Mar 10

Example, pre-train (reward free) to map temporal distances into distances in latent space, and then, finetune: map these through a dot product with a latent task description to a reward function.

A couple of refs:

openreview.net/forum?id=YGh...
arxiv.org/abs/2110.02719
arxiv.org/abs/2110.15191

1 2 4

Jonathan Balloch @balloch.bsky.social · Mar 10

I know exactly what you mean. Especially for us academic-related folks, our recommended bubble gets ultra tight. my recommendation is to look at some of the "highly followed" topics, which will give a more norm-y feed. But truly BlueSky needs "Trending"

4

Jonathan Balloch @balloch.bsky.social · Mar 10

Depending on precision, that is a crazy price for 2 high quality 6-dof robot arms, to say nothing of them attached as one torso. If the price stays when people start building it you can be sure I'll be one. The Rethink Baxter is a lesson, cumulative error from backlash will be the important thing

1 3

Jonathan Balloch @balloch.bsky.social · Mar 10

agreed

Jonathan Balloch @balloch.bsky.social · Mar 10

very exciting!

Chris Paxton @cpaxton.bsky.social · Mar 10

$14k open source humanoid robot upper torso. Writing with a pen on a notebook that you're holding is an impressively challenging task! Also comes with an open, modular, python software stack for robot control and planning.

openpyro-a1.github.io

1

Reposted by Jonathan Balloch

Chris Paxton @cpaxton.bsky.social · Mar 10

$14k open source humanoid robot upper torso. Writing with a pen on a notebook that you're holding is an impressively challenging task! Also comes with an open, modular, python software stack for robot control and planning.

openpyro-a1.github.io

3 14 58

Reposted by Jonathan Balloch

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · Mar 10

Hiring researchers and engineers for a stealth, applied research company with a focus on RL x foundation models. Folks on the team already are leading RL / learning researchers. If you think you'd be good at the research needed to get things working in practice, email me

2 11 65

Jonathan Balloch @balloch.bsky.social · Mar 10

Begs the question: at what point is multi-task training implicit meta learning @chelseafinn.bsky.social

1

Jonathan Balloch @balloch.bsky.social · Mar 7

1

Jonathan Balloch @balloch.bsky.social · Mar 6

Congrats Andrew and Rich, well deserved!! apnews.com/article/turi...

AI pioneers who channeled 'hedonistic' machines win computer science's top prize

Teaching machines in the way that animal trainers mold the behavior of dogs or horses has been an important method for developing artificial intelligence and one that was recognized Wednesday with the...

apnews.com

3 6

Reposted by Jonathan Balloch

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · Feb 20

One reason to be intolerant of misleading hype in tech and science is that tolerating the small lies and deception is how you get tolerance of big lies

4 27 190

Jonathan Balloch @balloch.bsky.social · Feb 20

super excited to try this out

1

Reposted by Jonathan Balloch

Nathan Lambert @natolambert.bsky.social · Feb 13

Trying to tell the story behind this explosion of research we are in. An unexpected RL Renaissance.
New talk! Forecasting the Alpaca moment for reasoning models and why the new style of RL training is a far bigger deal than the emergence of RLHF.
YouTube: https://buff.ly/41bVRPp

An unexpected RL Renaissance

New talk! Forecasting the Alpaca moment for reasoning models and why the new style of RL training is a far bigger deal than the emergence of RLHF.

www.interconnects.ai

3 11 65

Reposted by Jonathan Balloch

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · Feb 20

Easier installation, faster PPO script, new tutorials. The team has put in so much work and I'm excited for y'all to try it.
github.com/Emerge-Lab/g...

1 2 29

Jonathan Balloch @balloch.bsky.social · Feb 20

Incredibly cool article. Why, in spite of all of the hype about the scale of learning, we shouldn't forget the second half of Sutton's Bitter Lesson: search scales too, and often better yellow-apartment-148.notion.site/AI-Search-Th...
(h/t klowrey)

AI Search: The Bitter-er Lesson | Notion

What if we could start automating AI research today? What if we didn’t have to wait for a 2030 supercluster to cure cancer? What if ASI was in the room with us already?

yellow-apartment-148.notion.site

2