This confirms my own intuition based off many self-experimentations.
How deep into sleep you go is probaly still the most underrated characteristic of health and recovery.
How well regulated & on schedule is your system such that it naturally triggers all the mechanisms for facillitating, deepening & maintaining "sleep".
There are so many angles to this given how so many mechanisms can be leverage here. But it has been shown consistently to me that pass a certain amount of both deep sleep as well as REM sleep in total quantity as well as a % of it meaningfully changes my entire state of being regardless of the total amount since you can spend most of it in a awake state.
I mentioned so many angles earlier, at a physiological level but surprisingly perception/psychology can play a big role, in that your environment and how consistent the triggers within it are, to faciliate sleep by convincing you sensorially.
It's smth that is so deeply interesting with the many levers to it. Human Biology is just so fkng goated
Does anyone actually sleep 8 hours/night?! I hear that thrown around a lot but I don’t think I know a single person who sleeps that much on a regular basis.
You know how hard it was for me to see OpenAI as the good guys compared to Anthropic ever sincr they betrayed their founding "Open Source" Value.
You know how distrutful one have got to be for me to cheer for OpenAI instead of Anthropic.
I would rather shoot myself twice in the head than trust a company that is the very representation of
"Believing that you are so fundamentally good so much so that you justify every one of your actiona"
"Believing so much in your version of "Morality" and staying all and almighty"
If a company like anthropic wins, it will be the absolute worst thing for humanity because of their performative "morality". All of it is about showing meaning if their was a path that led to less harm by AI but did not portray them as the one doing the most for doing so then they will not choose it.
THE VERY INCENTIVE THAT IS DRIVING THEM IS THE EXACT THING THAT WILL LEAD MISALIGNED AI IRONICALLY.
"the idea that there is such a thing as a human or a group that can behold the "best sets" of morality to a being such that they are devoided of bias? lol
yeah sure keep RLHF ur way into EVIL AI
@zohranmamd5n0w@YashHustle_22 bruh not even that long ago ChatGPT at Usage limits and u didn't even fucking have usage %. It was this amount of message per hours like 40 and ppl would cram everything into one promp.
U guys are talking as if that shit is a new thing or smth
@ivicask@YashHustle_22 u guys werent there from day 1 & u are taking shit for granted but chatgpt started exactly where codex is at and it was way worse then that.
Can't believe how entitled ppl have become when ChatGPT would just randomly after a couple msg tell u: "Limit resets tomorrow..."
Yes, if you use Clerk for auth. However, the real issue with Cloudflare is the DX with a lack of a unified dashboard, and the UI/UX not being as clean or being a bit overwhelming if you are new to it.
Choose your poison, yet honestly, I believe it's 1000% worth it, and I do feel like I get to understand more about infra
@BTCqzy1@LexnLin Be proud of your work and I am willing to bet a band that at the very least the @OpenAI team might want to work w u or smth.
Bruh has the sauce to make GPTs actually good at front end somehow 😭
Work with him or smth @thsottiaux
@LexnLin bruh they should just hire you at this point😭
Talking about "cracking" it... why can't they just get you to build w the thing internally atp😭
@SIGKITTEN could u slip a word or smth
cc: @thsottiaux@ajambrosino
What I have been on the verge of doing is to somehow replicate the capabilities of ChatGPT through a series of MCPs, Integrations with web search capabilities and related tools all bundled into one Plugin, yet it wouldn't be seamless UI/UX wise.
I would still need to manually do projectless chats...
I have been having a very similar issue workflow.
I essentially use Codex as the core to any execution, yet preparing the execution is a crucial part which requires deep research, planning and synthesis so ChatGPT.
If these 2 truly merged, I don't think I would ever leave OpenAI's ecosystem...lol
@cs_serdar@willdepue w rollbacks & branching from rollbacks and stuff??
ish... not sure about that
This would massively blow u compute resources. Remember that it's a different scale when u are working with local data, codebases and stuff.
@LexnLin they had it before I believe unless i hadn't seen the cursor one but this post is actually after the fact that it was already in my interface for at least the past 2-3? days
Technically, there is still a lot of value in having an AGENT/.md as the core instruction file aka Ultimate System Prompt that influences everything else given that skills, plugins are "on demand" non persistent"
For those that are confused with the difference between plugins and skills.
One simple mental framing that i have now is to refer to
CORE = AGENT/.MD
SKILLS = Reusable "Task" specific capabilities or "Workflow"
PLUGINS = Reusable Bundle of Skills, Toolings & Integrations
We’re making Codex more useful for your work by expanding plugins beyond individual tools.
These plugins turn Codex into a specialist for a specific role with a single install, no coding required.
Codex can access 62 popular apps and 110 skills for work across sales, data analytics, creative production, product design, and public equity investing.
https://t.co/nunrYP2uMI
It's mind blowing how quickly we went from cumbersome to manage .md skill files to a fully integrated system but not only that, but with the plugin being a kind of context specific sandbox env w all the tools, integrations & more that that context needs.