Sabitlenmiş Tweet
Plugyawn
1.2K posts
Plugyawn
@plugyawn
behold the chorus of the civilizational accelerando.
Katılım Ekim 2022
466 Takip Edilen260 Takipçiler
Delve accused of misleading customers with ‘fake compliance’ techcrunch.com/2026/03/21/del…
English
@plugyawn @HassanRIsmail Cross product is high school stuff lmao what
English
Something I find strange is the large number of 'AI labs' being built by 1st or 2nd year undergraduate dropouts.
Statistics (big one), linear algebra, and calculus - all of which are literally the core of ML, and especially RL. Are all dropouts really these 'hidden' academics?
English
Plugyawn retweetledi
probably a good idea to gift your parents a subscription.
nexxel@nexxeln
if you’re in India, subscribing to @opencode just got a lot easier upi autopay is now live for opencode go ₹900/month for generous limits and reliable access to open source coding models
English
on the context of musk suspiciously tweeting about smoking a lot recently, i heard someone say that if someone truly believes something like @IsomorphicLabs will solve all disease, then it is the accelerationist skin-in-the-game decision to smoke.
Elon Musk@elonmusk
You can Chibify anything 😂
English
Plugyawn retweetledi
Plugyawn retweetledi
English
We've reached an agreement to acquire Astral.
After we close, OpenAI plans for @astral_sh to join our Codex team, with a continued focus on building great tools and advancing the shared mission of making developers more productive.
openai.com/index/openai-t…
English
@_TarunKathuria @DanielleFong damn thanks so long looking at schrodinger bridges on AISTATS listings and i've never come upon this. this is great.
English
Actually diffusion models are just the Schrödinger equation with time being real vs complex in the latter. leonard.perso.math.cnrs.fr/papers/2014-Le… ( diffusion models are referred to as Schrödinger bridge- I’m being a bit imprecise since Schrödinger bridge is between two general distributions whereas diffusion models assume one is the stationary distribution of the underlying Markov process). Flow matching is much simpler: your coupling is simpler and is first defined between two specific points in the pair of distributions and then a deterministic velocity field is learned over the uniform coupling between these two distributions)
English
@anirudhg9119 Funny how the first sequence attention paper was also a good ol' Bengio.
English
We (with Alex lamb) tried selectively aggregating information from all previous layers (instead of using residual from the past layer).
arxiv.org/abs/2010.08012
Kimi.ai@Kimi_Moonshot
Introducing 𝑨𝒕𝒕𝒆𝒏𝒕𝒊𝒐𝒏 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒔: Rethinking depth-wise aggregation. Residual connections have long relied on fixed, uniform accumulation. Inspired by the duality of time and depth, we introduce Attention Residuals, replacing standard depth-wise recurrence with learned, input-dependent attention over preceding layers. 🔹 Enables networks to selectively retrieve past representations, naturally mitigating dilution and hidden-state growth. 🔹 Introduces Block AttnRes, partitioning layers into compressed blocks to make cross-layer attention practical at scale. 🔹 Serves as an efficient drop-in replacement, demonstrating a 1.25x compute advantage with negligible (<2%) inference latency overhead. 🔹 Validated on the Kimi Linear architecture (48B total, 3B activated parameters), delivering consistent downstream performance gains. 🔗Full report: github.com/MoonshotAI/Att…
English
English
@neeldhara @kshashi @CompSciFact I don’t know if I want to impose my humour on an institutional account and get it banned.
English
What kind of content do you expect/look forward to from a CS/E department handle?
English
"guys our new model's a GDN!"
Claude@claudeai
1 million context window: Now generally available for Claude Opus 4.6 and Claude Sonnet 4.6.
English