HNHacker News
TopNewBestAskShowJobs

tempaccount420

767 karma · joined November 23, 2022

submissionscomments
tempaccount420··on Can we talk to whales?
I don't think parent is talking about you then? You have an inner monologue.
tempaccount420··on Beating GPT-4 on HumanEval with a fine-tuned CodeLlama-34B
Slippery slope
tempaccount420··on GPT-3.5 Turbo fine-tuning and API updates
I wouldn't be surprised if students will soon adopt the style of ChatGPT, say because they're using it to learn English, or just talking to it a lot.
tempaccount420··on Changing its name tanked X's downloads in App Store and Play Store
> there would be thousands of detransitioners already

https://www.reddit.com/r/detrans

tempaccount420··on Changing its name tanked X's downloads in App Store and Play Store
How? Trans"sex"uals are absolutely about sex, not gender.
tempaccount420··on GNU Parallel, where have you been all my life?
Why not just cite it?
tempaccount420··on I'm 17 and wrote this guide on how CPUs run programs
Can't wait for people to start abusing this strat by lying about their age.
tempaccount420··on The BBC on Mastodon
Federations are the worst way to implement decentralization. All because people are afraid of blockchains.
tempaccount420··on Jupyter Notebook 7
Same here - VS Code or Google Colab if I need an Nvidia GPU. I wish I could get Google Colab's GPU in VS Code, like Paperspace lets you do: https://docs.paperspace.com/gradient/notebooks/notebooks-rem....
tempaccount420··on Zenbleed
Thanks, that's what I meant.
tempaccount420··on Zenbleed
That's not very convenient. I want docker to be able to use all my cores when I build an image.
tempaccount420··on Firefox has surpassed Chrome on Speedometer
Firefox is great, Mozilla on the other hand...
tempaccount420··on Orca rams into yacht near Scotland, suggesting the behavior may be spreading
Hopefully no one tries to find out. There's only 50k orcas in the world, they're more valuable than us.
tempaccount420··on Zluda: Run CUDA code on Intel GPUs, unmodified
Although true, I don't think we should be trying to circumvent his block.
tempaccount420··on Has ChatGPT Been Neutered?
It seems pretty logical to me. Fine-tuning to make it more polite is giving it questions and punishing for giving an actual answer.
tempaccount420··on Tell HN: We should start to add “ai.txt” as we do for “robots.txt”
I wouldn't want to be you when Roko's Basilisk emerges.
tempaccount420··on It looks like GPT-4-32k is rolling out
As far as I know it's not documented anywhere and there is no way to ask the team at ChatGPT questions. I sent them an email about it a few days after GPT-4 release and still haven't received a reply.

Another thing that annoys me is how most updates don't get a changelog entry. For whatever reason, they keep little secrets like that.

tempaccount420··on It looks like GPT-4-32k is rolling out
> current 8k GPT-4 is already restricted to 25 prompts/3 hours

I'm pretty sure they're using a 4k GPT-4 model for ChatGPT Plus, even though they only announced 8k and 32k... It can't handle more than 4k of tokens (actually a little below that, starts ignoring your last few sentences if you get close). If you check developer tools, the request to an API /models endpoint says the limit for GPT-4 is 4096. It's very unfortunate.

tempaccount420··on sudo and su Being Rewritten in Rust for Memory Safety
That's not what they said.
tempaccount420··on Russia killed its tech industry
Yep. Same tricks that the CCP plays on it's population.
tempaccount420··on Show HN: WasmGPT – “ChatGPT” in the browser, no WebGPU and no server needed
This advice doesn't apply to ChatGPT in my experience. It does great either way, probably thanks to great RLHF.
tempaccount420··on Tell HN: ChatGPT is fantastic for finding and solving issues in logs
That's API, ChatGPT is different.
tempaccount420··on Agents are the next AI killer app after ChatGPT
That sounds a lot like the broken window parable. https://en.wikipedia.org/wiki/Parable_of_the_broken_window
tempaccount420··on StableLM: A new open-source language model
StableLM already has an optimal parameter number to tokens ratio, so what's your point? They should train the 65B model on even more tokens?

> StableLM is trained on a new experimental dataset built on The Pile, but three times larger with 1.5 trillion tokens of content

tempaccount420··on StableLM: A new open-source language model
If you want it to just regurgitate training data, sure. But more parameters will always be better for more complex tasks.
tempaccount420··on Astral
The user expresses their frustration with the critics by sarcastically asking what it would take to impress them. They believe that the new technology offers significant benefits and that the complaints seem unjustified.
tempaccount420··on New macOS malware steals info, including a user's entire Keychain database
Do what mobile OS do and make the app ask for permissions just before it needs to do something special. Make it impossible to ask for root. Sandbox apps.

Linux distributions package applications (including binary ones) for their users, why can't macOS or Windows do this? Run automatic safety checks on them before deploying updates to users. The Windows Store and Apple's App Store (on macOS) are kind of like that, but they only have applications that developers chose to put there. I don't see Linux distributions asking developers for permission to package their application (there are proprietary packages on Arch and NixOS), so I don't see why Apple couldn't do the same.

Certificates are a bureaucratic half-solution.

tempaccount420··on New macOS malware steals info, including a user's entire Keychain database
To sign your software you need to dox yourself to your users and pay a fee for a cert. I think there should be organizations that will sign open source software for you with their certificate, not sure if this is allowed.
tempaccount420··on GM’s decision to ditch Apple CarPlay, Android Auto sparks backlash
Makes texting less comfortable though IMO.
tempaccount420··on The inner beauty of a Tesla Stropkov DS 3600 telephone from the early 1980s
"useless complexity," spoken like a true software engineer.
← PreviousPage 10 of 12Next →