Rex Ledesma

@poolsideai @NVIDIAAI it's a beauty

Turns out, the next Laguna also fits on a @NVIDIAAI DGX Spark 👀

i am back on my grind to challenger as your favorite one trick

@sidpshanker congrats sid!!

working at baseten has been a crazy ride, have packed a lot in to the last few years. but we're just getting started 👀

@Sauers_ it scored very high on our whimsy eval

M.1

Poolside Summer is just starting 🏖️

@thdxr lots of cope unfortunately, but I'd like to say that it's great for innovation. very ripe for creative destruction x.com/rexrledesma/st…

this dynamic is really crazy right now even things for building other things are getting wrapped with another layer trying to be the access point everyone is flailing, full of anxiety, zero restraint

sadly, it is a form of cope. many people want a semblance of control as capabilities collapse into the models

@charliermarsh i love it too, it's so great for centralized control x.com/rexrledesma/st…

I love ast-grep

@fleetingbits reminds me a lot of nadia’s idea machines

some thoughts on the difference between oai and anth 1) i think the most important difference between openai and anthropic is cultural; basically, openai is faang while anthropic is ea 2) openai doesn't really believe anything; it's a lot of tech workers from meta, etc... most of the employees are not particularly agi-pilled 3) anthropic has roots in the ea ecosystem; it employs holden karnofsky, joe carlsmith, amanda askell, etc... all deeply connected to effective altruism 4) the founders essentially took an ea pledge; the series a was led by sbf, the series b was led by jaan tallinn (ea or ea adjacent) 5) this gives anthropic a core set of values and an orientation towards the future; it inherits effective altruism's focus on forecasting and ai safety 6) and, it gives it a set of institutions to recruit from and to reinforce its values: open philanthropy, mats, metr, constellation, redwood research, etc... 7) not that these organizations are all ea in particular, but they are part of a larger ea and ea funded ecosystem that curates a particular worldview and set of attitudes 8) openai has no equivalent; there is no concentrated place from which it draws talent or from which it draws a coherent worldview; big tech isn't exactly a reservoir of ideas 9) i think this may be one of the reasons that openai fumbled with their focus on ads and sora rather than focusing on automating the office worker; they were not future focused enough 10) and, it's one of the reasons why it is easier for anthropic to retain talent, there are a lot of voices telling the people working there they are doing the right thing and the most important thing 11) but, believing in nothing in particular, other than the zeitgeist, has its benefits, it can be easier to make deals (see openai's dod deal); and believing in things can make it harder 12) and, this tendency may make openai more responsive to the general political milieu and will probably make it more likely to try to follow the democratic policy discussions rather than lead them 13) and, it is an open question whether we want companies to lead policy making on ai in democratic countries or whether it is better to allow broader society to do this

slow, steady, then all at once 😍

does anyone know if this is a good thing?

please try to enjoy each token equally, and not show preference for any over the others.

Laguna M.1 and XS.2 now support 256K context. Laguna M.1 is now live with a 256K context window on the Poolside API and OpenRouter. With this update, it reaches 45.8% on Terminal-Bench 2.0, improving long-horizon performance. Laguna XS.2 is also moving to 256K today, with the updated config already available on Hugging Face. Both models remain free to use. Over 1T tokens have been processed since launch 4 weeks ago. Excited to see what people build with the longer context window.

Check out the full Laguna technical report! We talk more about our magical Model Factory system that powers @poolsideai :) x.com/poolsideai/sta…

dagster mentioned

Today we’re publishing the technical report behind Laguna M.1 and Laguna XS.2. This report opens up more of what went into them: Model Factory, pre-training data, distributed training, post-training, agent RL, quantization, and evaluation. poolside.ai/assets/laguna/…

On this day, last year. Thank you to @recursecenter for all the energy you've given to my life ❤️

@gabriberton This will soon be handled at the type checker level. With pyrefly, you can enforce these as static type checks rather than enforcing it only at runtime through assert or jaxtyping pyrefly.org/en/docs/tensor…

Python code should be full of asserts Duck typing is just dangerous for deep learning: pass a list instead of a tensor, and you get a weird bug instead of "pass a tensor, not a list" If the code needs a tensor, add an assert

Just published my notes from this year's Python Typing Summit at PyCon US. If you've ever been curious what one of these is like from the inside: intersection types, constraint sets in ty, tensor shapes in Pyrefly, Guido on direction. 🐍📝 bernat.tech/posts/pycon-us… #PyConUS

visiting my french sibling coop 🐺