LLMaxxing @nvidia ; Design is not finished. Common prosperity. Prev @poolsideai
LLMaxxing
Joined September 2016
- Tweets1.5K
- Following1.3K
- Followers2K
- Likes72.9K
Eric Alcaide retweeted
Look at him, thats my Qual.
Your what?
My Qualitative! My vibe specialist. Look at him, you notice anything different about him? Look at his face.
I'll give you a hint, his name is Jayden. He won a local poetry competition in bushwick! Yeah I'm sure of the vibe.
Quant firms are hiring idea guys, not quants
I've been skeptical about the whole - software is commoditized, ideas are the moat argument
But I've seen first hand the best quant and technical firms in the world hire thinkers, liberal arts grads, and generalists
I predict the trend accelerates
Letter from @AQRCapital below
Eric Alcaide retweeted
A short manual on how to systematically increase your exposure to opportunity.
(1/2)
Eric Alcaide retweeted
Readers added context they thought people might want to know
The video is an animation by artist R.J. (@RJ16848519) from his Screen Time series, shared here without credit.
x.com/rektober/statu…
x.com/connoissurreal…
x.com/kikanicolela/s…
Eric Alcaide retweeted
Making RL environments less broken and unfair to models seems to be extremely underrated as a strategy for improving alignment.
Eric Alcaide retweeted
Gràcies Fundació Bofill, ensenyament per projectes, aula invertida, coavaluació, TIC a les aules, fes-me un dibuixet, no memoritzeu, fora els llibres, visca les pantalles, IA, reels i tik tok.
⚠️ #ÚltimaHora Caiguda sense precedents a PISA: Espanya retrocedeix l'equivalent a un curs en comprensió lectora 3cat.cat/3catinfo/caiguda-se…
3cat.cat/3catinfo/caiguda-se…
Eric Alcaide retweeted
Justo esto lo he vivido esta semana.
Yo iba agarrándole por el sillín y le solté.
Seguí corriendo a su lado, con la mano por detrás.
Le iba preguntando: ¿Te suelto? ¿Quieres que te suelte?
Y él me decía que no, que no le soltase... sin saber que ya le había soltado.
Cuando me paré y le dije que ya llevaba un rato pedaleando sin que le agarrase se le iluminó la cara y se le escapó la risa.
Yo, obviamente, me derretí.
Tened hijos, no seáis subnormales.
Eric Alcaide retweeted
It turns out that adding artificial rescaling symmetries to neural networks makes them sparse. In our 2022 ICML work, we showed that because rescaling symmetry + weight decay is equivalent to L1 sparsity, one can leverage this to train very sparse neural models.
arxiv.org/abs/2210.01212
Eric Alcaide retweeted
give them new sandboxes??
what do you even mean?
can ppl pls stop freaking out about models escaping their sandbox and just put a judge/classifier into their interception server/llm proxy, monitoring traces like any sane research engineer would?
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Eric Alcaide retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
Eric Alcaide retweeted
Dataset work is one of the most impactful work you can do in machine leaning and yet one of the least socially rewarded. Huge shout out to all the incredible people who made this happen <3
For over two years, the largest open code dataset was The Stack v2… until today.
🥞 The Stack v3 is out: the largest open code dataset ever released: 114 TB, 770 languages, 224M repositories, ~5T tokens of deduplicated and filtered source code. Fully open, no restrictively licensed code included.
The upgrade:
- v2 (2024): 68 TB raw -> 2 TB / ~550B tokens, 618 languages filtered
- v3 (2026): 114 TB raw -> 15.9 TB / ~5T tokens, 713 languages filtered
C++ x15, TypeScript x7.5, Rust x7, Python x4.8. Even the COBOL corner of GitHub got a bigger slice.
Part of that is two fresh years of open source. Part of it is a bug we found in v2's deduplication - story below. 👀
Things that make v3 different:
1. Contents inline. The #1 complaint about v2 was "cool dataset, where's the actual code?" v2 shipped file IDs from the Software Heritage graph, and fetching contents was a DIY treasure hunt. v3 is self-contained: sources embedded directly, one row = one repository. Download finishes -> you start training.
2. Fresh crawl. v2 was a 2023 snapshot of even older crawls. v3 is a direct re-crawl of GitHub at the latest commit, completed by August 2025: 224M repositories, 44B files. Forks were only included if they had 5+ stars.
Two ways in:
🥞 stack-v3-train - near-deduplicated, quality-filtered, PII-redacted, contents inline. Point load_dataset at it and go.
huggingface.co/datasets/Hugg…
🏗 stack-v3-full - the entire 114 TB corpus as an HF Storage Bucket: every duplicate kept with cluster IDs, stubs for excluded files. Roll your own dedup, filters, and mixes.
huggingface.co/buckets/Huggi…
The record isn't really ours to claim - it's millions of everyone's repositories, neatly stacked. Thanks to every developer who keeps their code public, the BigCode community, and friends who built the infrastructure that made this possible!
Eric Alcaide retweeted
This is a huge deal. An American open source model that I can run on my local machine and that performs on par with models 10X its size.
Today we're releasing Laguna S 2.1, our most capable model to date.
It's a 118B total parameter Mixture-of-Experts model with 8B activated per token, a context window of up to 1M tokens, and thinking and no-thinking modes.
Capable enough to hold its own against models many times its size. Small enough to run on a single @NVIDIAAI DGX Spark.
Laguna S 2.1 is fully open under OpenMDW-1.1, with weights available today on @huggingface
poolside.ai/blog/introducing…