Sharing an early result: Sierra verified Cortex at 73.2% on τ³-benchmark.
We build algorithmic complements to LLMs. We don’t fight what models do well. We fill in the rest.
https://t.co/MmTS0HVc3E
@FrancoisChauba1 Yes it seems everyone that builds any learning system calls it RSI as it became a fad. Meant to be AI improving itself causing science fiction singularities, but I guess labeling anything RSI if it is the latest valley fad probably gets attention nowadays
@FrancoisChauba1 Not saying parallel DFO/ES is a bad idea. We can and should do more, but you need a different plan of attack. If you want to use it for pretraining, change how you find the saddle point completely. I get the memory advantage, but directional derivatives to mimic GD may be hard!
@FrancoisChauba1@FrancoisChauba1 lets catch up in person dude - I see you are Chris Re's student, will be fun convo. If you have a gradient, use it or change the objective. If you don't, (like "model-free" RL), DFO/ES would be as good. Only objected to pre-train. Sure Duchi would agree.
BUYER BEWARE
This extremely meticulous article will now be used to start Phase2 of “shut down open source” because “if we can’t control our closed models, imagine what happens to everyday life when anyone can just allow their open source models to proliferate”.
Some will use this example and point to future grid outages, utility outages, transportation chaos, bioweapons, uncontrolled cyber swarms and the like to try and make the case for yet another attempt to limit US AI model control to a few folks.
Then, one layer beneath the propaganda, one sees that these things are always written and promoted by a web of well organized existing shareholders or shareholder-adjacent of the closed frontier labs and groups of folks organized by a shared ideology roughly rooted in them knowing best.
This article, more than anything else, should demarcate the beginning of a well organized effort to stand up a Closed Model Industrial Complex. These cartels exist in other markets as well and it now exists in AI.
Bottom line, read everything carefully before signing up for the consequences.
How we react will define the next hundred generations of Americans and their lives.
Do not give up your rights or sovereignty because of fear of the unknown from articles that, at the root, are as much about the hidden incentives of an emergent cartel and their proxies.
If you take the time to understand the incentives, you will be less likely to fall for it and overreact and have intelligence metered from a few suppliers.
Bottom line: Keep Calm. Carry On.
@thsottiaux 😂
Sam: "OpenAI should be a platform, not a product company" -> Bevy of products
Dario: "We must pace the frontier" -> Opus and Sonnet 5.5
https://t.co/PClxJINe2C
Sam Altman thinks OpenAI should be a platform, not a product company:
"What I think most people want is the single interface to their own personal or their company's AGI that can kinda help them with whatever they need, and then the ability with an API to build anything they want on top of it."
"That is the platform that we should offer to the world."
"I don't think we should go build every product category. I don't think we should go try to compete with all of our customers. I don't think we should try to subsume the entire economy."
"I think we should offer this platform and try to have a hundred million new businesses and eight billion people use it in all kinds of new ways."
"So one kind of direct interface to the product, one API for people to use however they want. Those eventually come more and more together too, and then it's all about what people do with it, build on top of it."
@sama w/ @davidsenra
@thsottiaux@thsottiaux This is worse news than Opus 5.5 for codex Tibo, hope DevDay makes up with something dramatic for codex. Consider more banked resets to ease the pain :)
@yoavgo Neither the paper, nor the currently misquoted FAQ seem to disagree with you on utility! At worst, misses the eloquence of https://t.co/OGKVIeTZJY Funnily, you'll see the renewed attack is by folks who wants to stop AI while this paper did the same for different reasons 😅
@yoavgo From what I have heard, Lean to verify occasionally that it was not a random rambling that is correct. Could very well have post trained as well. Lean is a trust-me-bro I'm not drunk today proof. LLM's doesn't deterministically backtrack, so they bruteforce or intervene