HN user

catasaurus

31 karma

some full stack and some ml dev

sebastianlarson22@gmail.com

Posts8
Comments26
View on HN

ray looks interesting, will check out

EDIT: so the main thing is technologies like Ray have a way to do these things, but I honestly just want an easy way to do this. Maybe means I will have to set up something with Ray and AWS myself and make a wrapper for that?

Yeah, I understand that it is something that can be done fairly easily but I am too lazy to do so at the moment (lol). By finicky I mean like usually kind of buggy and hard-to-use web interfaces to set up some weird Jupyter Notebook that they run or SSH (hard-to-use as in something like AWS Sagemaker). I understand the use of these things but I just want to be able to use my own development environment to do everything I need to do and then send some code over to run. Preferably the pricing would be kind of serverless, just paying for the time of the compute used to run the programs that are run, not just paying for the reserved server to sit there.

The paid version probably includes better servers and hardware for your ChatGPT instance, but OpenAI intentionally creating demand errors? I would think its more along the lines of they are using cheap servers to allow everyone to use ChatGPT and thus when too many people are trying to use this massive machine learning model (175B parameters), the wait time goes up and up as the models have more and more inputs to respond to.

Also most freemium platforms do try to motivate their users to upgrade to paid plans.

Spotify: No downloading music without paid plan, etc

AWS: Free tier but then you have to pay

Youtube: Ads but if you pay no ads

And on and on

Oh, reading this again. Do you mean something ChatGPT is excellent at answering? The only problem with that is that ChatGPT is really just an excellent bullshit machine, as in it is really good at making stuff up. So I would probably get some questionable information when it comes to smaller details.

Space begins at an elevation of 330,000 feet. For some reason, anything above this has been classified again, for some reason, as “outer space shall be free for exploration and use by all States;”. Below this, is when airspace laws and treaties begin to apply. Over US airspace, there are laws in place to allow commercial planes, and such from other countries to fly over. But for things like spy balloons, not so. It is considered invading sovereign airspace, and a violation of international law.

Overpriced compared to nothing. But for a chip that was released 3 years ago, maybe it is a little bit too much? Also yes, Nvidia does have a moat with CUDA, and it would take a long time for Intel and AMD to overtake them. So for the short term, Nvidia is probably fine just charging a lot of money for their chips. But in the end, if AMD or Intel releases a chip that competes with the A100 (or any flagship chip that Nvidia currently has a monopoly on), for a lower price, it is only a matter of time before high performance computing moves to the hardware interface of choice for said company. AMD released GPUFORT two years back, an alternative to CUDA for AMD products. Nvidia can afford to be doing what it is right now because of its monopoly. But that won't always be there. I don't have a firm opinion on what they should do, I am just trying to make sense of this.