@CtrlAltDwayne No one is old enough to remember gpt-4 that thing is ancient
English
Tech
6.8K posts
Tech
@tech_bag_
Buildooor. In it for the tech, left holding the bag.
Katılım Ocak 2013
3.3K Takip Edilen1.6K Takipçiler
Don't get too excited. Microsoft has an undefeated streak of taking powerful models and completely botching its own implementations. Anyone else old enough to remember them messing up GPT-4?
🚨 AI News | TestingCatalog@testingcatalog
Microsoft is exploring a deal to acquire Inception, an AI lab that recently released Mercury 2, the world's first reasoning diffusion LLM, according to Reuters. Imagine diffusion Copilot 👀
English
We’re gonna need a bigger boat
Tibo@thsottiaux
They meant companies... 2000 companies. We're going to need to stand up a lot of compute 😅
English
@rezoundous Always. Also just started spawning 5.3-codex-spark for implementation of discrete tasks with 5.5xhigh/fast as main orchestrator and it has been amazing. Don’t know why I wasn’t doing that before
English
Today we release Token Superposition Training (TST), a modification to the standard LLM pretraining loop that produces a 2-3× wall-clock speedup at matched FLOPs without changing the model architecture, optimizer, tokenizer, or training data.
During the first third of training, the model reads and predicts contiguous bags of tokens, averaging their embeddings on the input side and predicting the next bag with a modified cross-entropy on the output side. For the remainder of the run, it trains normally on next-token prediction. The inference-time model is identical to one produced by conventional pretraining.
Validated at 270M, 600M, and 3B dense scales, and at 10B-A1B MoE.
The work on TST was led by @bloc97_, @gigant_theo, and @theemozilla.
English
@MtnTimeInvestor @levie And it’s not just technically complex to deploy value add agents, but the true complexity actually lies in the particular enterprise and particular process. Which makes me curious how effective the FDE model will actually be.
English
I work for a big non-tech company, and can confirm technical talent is needed to build the agentic systems that leadership imagines. The idea of non-technical people vibe-coding applications is laughable without engineers that understand the existing IT infrastructure to build systems that actually drive outcomes. Otherwise its just a mess.
Today's enterprise leaders are pressured to demonstrate that their teams are using agents, but are not as interested in how they are being used. I think its similar to the .com era when we thought that just because you had a website, you were in the future. But a website isn't actually that valuable unless it can help you do something.
That's why tokenmaxxing can easily occur without resulting in any business value, and we're learning that its difficult to build agentic systems that can consistently deliver high-quality outcomes (slides created, content summarized, contract drafted etc.).
English
Forward deployed engineers, or equivalent, are about to become one of the most in-demand jobs in tech. And one of the most important functions for AI rollouts.
Deploying agents is far more technical of a task than most people realize, often far more involved than deploying software. Software generally works the same way every time, and generally for the past few decades has been updated versions of an existing technology or concept (which basically means easier for the enterprise to update their workflows on a newer system).
With agents, you’re actually deploying the equivalent of work output within the enterprise. The customer is effectively using you as a professional services provider for a task, which they expect to get solved nearly end-to-end now. This means you need to actually deeply understand the business process as a vendor, and get the customer from the current to the end state seamlessly.
Companies need help figuring out which models will work best for their workflows, they need extensive evals setup often, they need change management support for workflows, they need to get their data setup for the agents, and constant tuning of the agentic system for their process.
Massive role in tech now. And another example of the kind of highly technical work that AI is creating.
First Squawk@FirstSquawk
GOOGLE TO RECRUIT HUNDREDS OF ENGINEERS TO ASSIST CLIENTS IN EMBRACING ITS AI – THE INFORMATION
English
@thsottiaux @ajambrosino “Thoughts on collapsing releases to thursdays?”
“Yeah sounds good!”
“Dope. Will do. New release cadence is now Thursday (and probably tuesdays and a little bit of Friday”
😂🙌🏼🫶
English
Thanks for all the input, what we'll do going forward for Codex updates
- Tue: Quality and polish
- Thu: Big launches (starting this Thu)
- Fri: Fun stuff, little extra
Tibo@thsottiaux
For Codex, we’ve been thinking about keeping a stable release cadence and have a larger release each week on Thursday. That does make the start of the week and bit less exciting. Thoughts?
English
Tech retweetledi
I'm curious what folks think 'artifacts' means? What would be a better title for this section?
English
@jxnlco @ryanvogel @OpenAI Turns out the “you can just do things” also extends to “you should just do things”
Tech@tech_bag_
Massive consolation prize. As a thank you to @thsottiaux I made a little tiblet in his honor
English
Massive consolation prize. As a thank you to @thsottiaux I made a little tiblet in his honor
English
@ajambrosino @AishwaryaDevv True. But I filled half my context on a new session just “catching up” 😬 it’s getting there but not quite solved
English
@AishwaryaDevv if you use the codex app with memory on, you actually don’t have this problem
English
The GPT 5.5 party email coming at 5:55 was a nice touch 🤭
English
Feels like most people are missing his point. Of course “good” code still needs to be produced. It’s not more for the sake of more, it’s iterate much faster for the sake of solving real business problems. Yes first then true yes/no after is actually grounded in something tangible
Jonathan Ross@JonathanRoss321
For 50 years, software engineering ran on code rationing. Writing code was expensive, so we rationed it carefully through roadmaps, RFCs, prioritization meetings, and scope reviews. This created a role: the No Engineer. No, that won't scale. No, we don't have bandwidth. No, that's out of scope. No, we need a design doc first. The No Engineer was valuable for 50 years. Every "no" saved real money. Their judgment was the rationing system. LLMs will be the end of code rationing. Code is cheap now. And while the No Engineer is explaining why something can't be done, the Yes Engineer has already shipped three versions of it. If you're a Yes Engineer, the next decade is yours.
English
introducing the Codex Copilot Cowork Collaborative Connector Collection
English
This thing does more than what you think it does. Codex now available for non-coders
OpenAI@OpenAI
It's never been easier to do everyday work with Codex. Choose your role, connect the apps you use every day, and try suggested prompts. Codex helps with everything from research and planning to docs, slides, spreadsheets, and more.
English