foxhop
foxhop on Hacker News
Posted by
foxhop on Sunday March 23, 2025 @07:59AM
from the russell.ballestrini.net dept.
It is impossible to figure this out and it's dumb to try to give a number. We don't think Neanderthals could speak? They also had languages.
You scam artists are not fit for civilized society.
Politically motivated bollocks.
The white colonizer did not get colonized by tribes in Africa.
For somebody who cares about software why are you advertising for disqus comments? Link in bio to a public domain commenting system forn static sites.
When i buy 512G microsd it becomes my property. If i need to repair or replace my property, it should be disclosed what technology i bought, so that i or my repairman could understand my property enough to decide a proper path forward depending on my data & other environmental devices.
The author needed to reverse engineer what could have been on the spec sheet...
`tree --gitignore && cat .py && cat templates/`
I don't like lawyers so I use public domain.
Also code is like math which is public domain.
Posted by
foxhop on Tuesday January 07, 2025 @10:53AM
from the git.unturf.com dept.
Step 1: curate a context window of code from different repos (poke team about switching to mono repo)
Step 2: write a slack style message as if you are discussing the solution with a teammate that you have authority over as a delegate to get shit done & to revise as needed.
Step 3: press enter, LLM does something you don't like, delete history, fix prompt in step 2 and ask again, rinse and repeat until you have working code.
Step 4: ask for the changes to be written as a bash file that cat EOF all the files that change into place, run the script.
Step 5: git diff & play test the changes using functional testing (use your mouse & keyboard test the code paths that changed...)
Step 6: continue prompting & deleting history as needed to refine.
Step 7: commit code to repos
I was just thinking about something like this for very small web applications (1,200 line app.py & 600 lines of html templates), something lighter than requiring a working docker install.
a handful of dependencies (mostly Pyramid based at the moment) & whatever dependencies those have, pull it all down & serve it out of a tarball or zip file of a portable virtualenv.
If the maintainers are reading this is cool. I would start with an exporter/importer for pages from Wikipedia format into whatever format you use, it should also deal with the media. This is no small task but what you need.
Posted by
foxhop on Tuesday October 22, 2024 @10:11AM
from the russell.ballestrini.net dept.
the proxy/server accepts any API key. I expect people to copy this infra with their own domains, this is a working proof of concept.
Posted by
foxhop on Sunday October 20, 2024 @10:28AM
from the show-hn dept.
https://ai.unturf.com/#client-side
Client Side Only Example, chat with this page.Because we don't use API keys we don't have any real need for a server.view the page source, it's _all_ there!The program doesn't take history into account & yet is has surprising coherence.View Source and start on line 258.
The llama 3.0, 3.1, & 3.2 all use the TikToken tokenizer which is the open source openai tokenizer.
4090 has 24G
So we really need ~40B or G model (two cards) or like a ~20B with some room for context window.
5090 has ??G - still unreleased
My guess is it uses the same vocabulary size as llama 3.1 which is 128,000 different tokens (words) to support many languages. Parameter count is less of an indicator of fitness than previously thought.
The universe is all about various spectrum and waves like sin. It's donuts and toroids all the way up and down, left and right.
Here's a hint, STOP MINIFYING CODE! gzip over transport is enough.
Your going to need a lot more than a few, 800G VRAM needed
Thanks, the process to move off the platform is a process that needs to be built and documented, the data lives in a shared digital ocean bucket, the database uses uuids and we could try dumping only the relevant data out of the shared database. it was built as a multi tenet platform.
Future work is supporting api/secret keys to bring your own s3 bucket/digital ocean spaces, while keeping the database shared. otherwise it costs roughly $11/mo to host a copy of the platform ($6 for a virtual server and $5 for the digital ocean space)
As for taxes, this will be something that will need to be tackled at some point but so far I've been kicking the can down the road hoping I could find a partner to help figure that out in a clever way that is proper.
I built an app that lets creatives sell their digital and physical goods
https://www.makepostsell.com/
It's open source and public domain, uses stripe for holding card information
are there docs for this setup?