Players manage ship AIs to task sub-agents in a Factorio-like space trading environment. The open-source project utilizes Pipecat AI for backend, Supabase, and Vercel for the React frontend.
Information also reporting that Greg and others have left, or are about to leave. I wonder how long he's been gone, given he was still on Twitter promoting work until recently...
I wonder if that's true at a certain stage of OpenAI, which because of the product bootstrapping skills of Sam and co, has made his role irrelevant?
I mean, Jakub can take it forward at the current scale and leadership team of Sam and other people, but maybe he could not have earlier, which is where Ilya shone?
I don't have a horse in this race, and maybe the root comment came off as flippant and disparaging. But I'm not reading "outsiders" as being what you say "gatekeeping".
Maybe another perspective is that "outsiders" may not have the same view of the issue as experts in the field and may not (historically, in OP's experience) seem to want to work together with the experts to develop this view. Handwaving away complexities and not willing to get hands dirty is something I've seen as well so maybe I'm a bit more empathetic, but cold shoulders from "experts" towards newcomers is definitely a thing.
Both of which could help both sides - bring more depth to the fresh view of the "outsiders" and actually bring valuable freshness to the depth of the "experts".
I'm not sure folks who're putting out strong takes based on this have read this paper.
This paper uses GPT-2 transformer scale, on sinusoidal data:
>We trained a decoder-only Transformer [7] model of GPT-2 scale implemented in the Jax based machine learning framework, Pax4 with 12 layers, 8 attention heads, and a 256-dimensional embedding space (9.5M parameters) as our base configuration [4].
> Building on previous work, we investigate this question in a controlled setting, where we study transformer models trained on sequences of (x,f(x)) pairs rather than natural language.
Nowhere near definitive or conclusive.
Not sure why this is news outside of the Twitter-techno-pseudo-academic-influencer bubble.
This makes a lot of sense at deep tech places that don't sell to enterprises where one can put their head down and just work on technical challenges. A similar place like Mongo DB or the hardware division of NVidia comes to mind. But if you're supposed to iterate with any input from more than 2 people that isn't highly objective (technical metrics that can correlate well with product satisfaction), communication and diffused collaboration is important and necessary. That doesn't mean it has to suck. But the Netflix model you described isn't applicable everywhere and shouldn't be blindly replicated.
Apart from the debate raging on about grabbing/absorbing more from a physical book than an ebook, I feel that I'm simply more attached to the physical aspect of the reading experience.
I've read ebooks when I've been unable to find the hard copies, but I almost always revert to the hard copies.