3un01a @_3un01a

applied epistemologist @plastic_labs plasticlabs.ai unknown Joined May 2025

Tweets

68
Followers

121
Following

172
Likes

130

Dwarkesh Patel @dwarkesh_sp

4 days ago

.@RichardSSutton, father of reinforcement learning, doesn’t think LLMs are bitter-lesson-pilled. My steel man of Richard’s position: we need some new architecture to enable continual (on-the-job) learning. And if we have continual learning, we don't need a special training…

223 482 4K 1.4M 3K

Download Video

vintro @vintrotweets

5 days ago

who wants to train a custom agentic reasoning model (think Kimi v2) with tool calls to @honchodotdev in the dataset that literally tune your model towards your userbase?

0 1 5 722 0

3un01a @_3un01a

2 weeks ago

Congrats on the ship @itsalltruffles! Can’t wait to pop this one open!

9 4 88 10K 15

Download Image

Vishal S Pandey @its_vayishu

2 weeks ago

What if I told you... Softmax attention ≈ solving an entropy-regularized optimal transport problem? Turns out, attention is just a transport plan with a negative dot product cost. Full derivation below, visuals speak louder than words.

25 58 791 40K 757

Download Image

Simone Scardapane @s_scardapane

2 weeks ago

*The Origins of Representation Manifolds in LLMs* by Modell et al. They study the presence of "interpretable features" in LLMs embedded as manifolds and how their geometry connects to the internal representations of the models. arxiv.org/abs/2505.18235

15 162 1K 52K 691

Download Image

3un01a @_3un01a

2 weeks ago

had the best week at plastic major things are cooking

0 0 3 75 0

Blaise Agüera (@blaiseaguera.bsky.social) @blaiseaguera

4 weeks ago

For those of us raised on science fiction, having AI become part of everyday life feels like the breakthrough we’ve been expecting. Yet some critics insist that no matter what these models do, they're not truly intelligent—just clever simulacra.

7 13 73 49K 25

Download Video

3un01a @_3un01a

3 weeks ago

let the simulation commence

0 0 4 114 0

3un01a @_3un01a

3 weeks ago

but isn’t language and other “generated” construct also part of what we consider nature as well? perhaps the separation/divide between man and nature makes it difficult to understand what memory and intelligence embodies…?

0 0 3 99 0

Rohan Paul @rohanpaul_ai

3 weeks ago

Fei-Fei Li (@drfeifei) on limitations of LLMs. "There's no language out there in nature. You don't go out in nature and there's words written in the sky for you.. There is a 3D world that follows laws of physics." Language is purely generated signal.

257 704 4K 1.8M 3K

Download Video

elvis @omarsar0

3 weeks ago

Emergent Hierarchical Reasoning in LLMs The paper argues that RL improves LLM reasoning via an emergent two-phase hierarchy. First, the model firms up low-level execution, then progress hinges on exploring high-level planning. More on this interesting analysis:

16 93 562 105K 564

Download Image

Thinking Machines @thinkymachines

3 weeks ago

Today Thinking Machines Lab is launching our research blog, Connectionism. Our first blog post is “Defeating Nondeterminism in LLM Inference” We believe that science is better when shared. Connectionism will cover topics as varied as our research is: from kernel numerics to…

240 1K 8K 3.2M 5K

Download Image

3un01a @_3un01a

3 weeks ago

wonder if memory some how has memetic properties…?

2 0 4 133 0

Jack Morris @jxmnop

3 weeks ago

curriculum learning will make a comeback. unfortunately, random sampling is not the way heard it here first

44 15 584 47K 161

Ryan Greenblatt @RyanPGreenblatt

4 weeks ago

x.com/i/article/1963…

3 12 213 43K 240

Franky @FrankYouChill

4 weeks ago

A 14B model just beat a 671B model on math reasoning. Here’s how Microsoft’s rStar2-Agent achieves frontier math performance in 1 week of RL training - by “thinking smarter, not longer.” 🧵

27 208 2K 206K 1K

Download Image

Erik Hoel @erikphoel

6 months ago

🚨 New paper, and big personal news! 🧵 First, I just published a new theory of emergence. It traverses the dimension reductions of a system, treating each scale like a 2D slice of a 3D object, looking for what each adds causally and irreducibly. arxiv.org/pdf/2503.13395

51 131 778 166K 664

Download Image

3un01a @_3un01a

4 weeks ago

ask away... pennyforyourthoughts.ai/?expert=3un01a

0 2 10 1K 0

Download Image

Gorgias @2reb_fl

a month ago

You can question Gorgias the sophist directly about the nature of the human condition! Super smooth work from @plastic_labs @honchodotdev. Monetize with x402 @CoinbaseDev

3 2 16 60K 3

Download Image

Plastic Labs @plastic_labs

a month ago

Fresh @honchodotdev + x402 (@CoinbaseDev) demo! Penny for Your Thoughts is a demo where anyone can share their expertise and sell bits of it via micro-transaction. Hop on, get interviewed, set your price, & collect payments!