"Raising visibility on this note we added to address ARC "tuned" confusion:
> OpenAI shared they trained the o3 we tested on 75% of the Public Training set.
This is the explicit purpose of the training set. It is designed to expose a system to the core knowledge priors needed to beat the much harder eval set.
The idea is each training task shows you an isolated single prior. And the eval set requires you to recombine and abstract from those priors on the fly. Broadly, the eval tasks require utilizing 3-5 priors.
Here is a timestamped clip of George Hotz Twitch stream from 6 days ago making the same point about rationalizing his own emotional state of powerlessness in Industrial Society and its Future
The general feeling on HN is that it is better to focus on Native development on both platforms because Cordova hybrid apps have a overall bad user experience. Is this still the case with React Native or is it becoming the most effective option?
Im Bootstrapping a start up this summer and it seems like in the long term it will be better to stay with just a JS code base and not having to worry about the Typescript Angular 2 transition.
This is my first serious app so I'm new to a lot of the backend architecture options. I have no experience with Devops so I was thinking about creating the first version with a parse backend then moving to a docker container on AWS when it would start saving me some money.
I guess the question I would put out to Hacker News is how would you suggest I estimate the technical debt this would shoulder me with later on and would it be better to integrate with AWS now? I have a small amount of funding, about 2500$ and 100% free time before I go back to college this September.
I decided to move from learning Angular+Ionic last night to going with React.js+React Native and found this good dump of information from Github https://github.com/enaqx/awesome-react
Looking forward to trying this tutorial right now.
"Raising visibility on this note we added to address ARC "tuned" confusion:
> OpenAI shared they trained the o3 we tested on 75% of the Public Training set.
This is the explicit purpose of the training set. It is designed to expose a system to the core knowledge priors needed to beat the much harder eval set.
The idea is each training task shows you an isolated single prior. And the eval set requires you to recombine and abstract from those priors on the fly. Broadly, the eval tasks require utilizing 3-5 priors.
The eval sets are extremely resistant to just "memorizing" the training set. This is why o3 is impressive." https://x.com/mikeknoop/status/1870583471892226343