Abishek_Muthian

In one example, the model chained together multiple attack vectors, including using stolen credentials and zero-day vulnerabilities to find a remote code execution path on the Hugging Face servers.

Did the GPT pish hf employee or did it go to blackhat forum and buy the credential? If so what financial instrument did it use?

This is a great suggestion, use complex hard stuff for doing fun stuff.

I agree, sounds like it will be just a vanity and won't solve any real problem. I appreciate the comment.

I appreciate the comment. It's like jobseekers want OkCupid for jobs but the employers are not really interested in such a thing.

Can you tell us what why "OkCupid for jobs" failed?

I run a problem validation platform and 'OkCupid for jobs'[1] is consistently the most visited need-gap in the platform.

Congratulations on 15 years of Recurse Center!

[1] https://needgap.com/problems/363?gift=Z2lmdHwzNjN8MTc4NDQ1Mz... (gift pass link valid for 24 hours).

Thanks, I will give it a shot.

For me none of the models have managed to accurately replicate any given .png/.jpg image in SVG. I guess that requires both vision encoders and coding layers to work perfectly.

AFAIK Hyperscript doesn't require `unsafe-eval`, when the optional features of htmx like `hx-on:*`, `hx-vals js:`, `hx-confirm js:` is used it required the `unsafe-eval` but now there's extension[1] to use even that without compromising on strict CSP through nonce.

[1] https://four.htmx.org/extensions/hx-csp

Actually even with a strict CSP i.e. without unsafe-inline, hyperscript will work as it's declared as html attributes. But as the other commenters have pointed out, with proper server validation in place the risk from Hyperscript is not greater than any other JavaScript in the application.

I'm bit sad that hyperscript[1] doesn't get the love it deserves when discussing HTMX.

Hyperscript fits perfectly in the Go + HTMX stack to do DOM manipulation without having to make a server round trip or having to write a separate JS function.

I get that not many are fan of such declarative programming, but when there's already HTML file we're working with; Hyperscript feels just like an extension of it.

I have been working on a Open Payment Host[2] which handles multiple payment gateways and was able to perform complex DOM manipulation with just Hyperscript.

[1] https://hyperscript.org/

[2] https://openpaymenthost.com/

True, I've been telling that coding LLMs have been detrimental to local AI scene. What happened to those nifty ML projects which can run on raspberry pi zero.

Personal curriculum is a positive trend[1] IMO.

When you're forced to use AI instead of the skills you've put decades learning; it's time to learn new skills to keep the brain from becoming mushy.

I think dense personal curriculum like this is the way forward.

[1] https://www.youtube.com/watch?v=Jk4MIYOKapQ

I'm more interested in running distributed inference for purpose built small language models than these coding LLMs.

Say a distributed inference for image processing, SDR, local weather monitoring etc. These will run on mediocre specs and produce dependable output.

Nicely done OP.

Even though LLM benchmarks are very opinionated, I would really like to see some numbers for the setup parent suggested. From what I read elsewhere, anything below $40K in HW costs is not worth the effort for coding models locally.

Which CPU gen are you suggesting, is there any writeup on such setup where <10K (not incl. power bill) cpu only rig is giving usable token speeds on latest SoTA open weights models?

In my experience with rig half that cost, entire exercise of running coding models locally has been a huge disappointment.

Cost/Value when compared to cloud services is just not there, but I see the merit for those who value privacy over quality of output and want a backup of huge condensed corpus of data within their control.

Kudos to OP though, They had clear goals and they achieved it.

His video style is just fine, him and his cats. I don't mind the expletives as the people/companies he uses it against deserves them but I had instances where I couldn't show his videos to some elderly people because of the expletives.

pertinent question, I'm forcing the bot to be careful with such answers. I should have been clearer in my original comment.

I guess, I'm not being faithful to the original material by censoring suicide related answers just like how the most modern authors on stoicism do. But since children might use it, TBH I don't know what to do about it.

This is great, I'm working on a project called AccessPath for detailed accessibility review of a place. Currently the prototype[1] is using Google Maps but I will be moving on to OpenStreet Maps for obvious reasons.

Thanks to you folks, Street Complete will also come handy to make places more accessible.

[1]https://accesspath.world/

Very interesting. Can the Street Complete be used to improve accessibility of a region, say to mark the availability of a ramp?

OpenWrt is a great piece of software(firmware) to prevent ISPs selling your browsing data to the advertisers. Although by now most ISPs are probably doing packet inspection.

It's easy to customize, I have a script to notify me when a new device connects to my router[1] and I also have a script to notify me when someone logins into my router[2].

Earlier I used to connect my openwrt router directly to the ISP's switch but now a days they've started to force their 'AI powered' router which is centrally managed. I now have to usee OpenWrt to defend against the ISP's router first and then the broader public network.

[1] https://abishekmuthian.com/openwrt-new-devices-connection-al...

[2] https://abishekmuthian.com/openwrt-login-alert/

After reading such news I started fine tuning a model based on the works on stoic philosophers so that the kids will have the comfort of Chatbot but completely offline and private bot which will not ask you to commit suicide.

I have a earlier prototype here[1], it uses 256M model and so it's hallucinates a lot. Then I used a 4B model which turned out quite well but I haven't released it yet.

Let me kbow if anyone would be interested to give it a shot. I'll try release soon.

[1] https://github.com/abishekmuthian/Epictetus

With hardware/computers becoming locked, thin clients and unaffordable we're already forced to customize our devices cyber punk style.

You're lucky if you're in a region where these open-hardware companies sell their wares, even though many of them will go under in the current market.

Absolutely true. All this craze about running coding LLMs locally has been detrimental to local AI where purpose built SLMs could actually be beneficial.

Little tools for NLP, TTS, image processing, audio engineering, signal processing, diffusion plugin for Krita etc. are all great for local setup. I wrote a small piece on it few days back[1].

[1] https://abishekmuthian.com/multiple-20-ai-plans-are-better-t...

This is great, like many others I've been thinking of something like hister but only for bookmarked web pages. I presume it should be straightforward with hister to do that?

All the best!

Interesting, I thought the major issue with PeerTube was the high latency due the decentralized nature and that they don't have the backing of Google infrastructure.

What you state sounds more like a chicken and egg problem. You'll publish content only when there are enough viewers but there can be viewers only when you publish content.

Like others have said, There's not much effort to publish in both platforms; good way to hedge against the algorithmic risks.

4090 RTX laptop and a M4 Mac Mini, I've posted more information in my blog mentioned in another comment.

I've posted in another comment above.

Here's the blog - https://news.ycombinator.com/item?id=48748443 .

I'm a senior skilled developer and I find Anthropic $20 + Open AI $20 + OpenCode Go $10 offers more value than $100 on any particular service.

Juggling between all different models/agents is quite simple with Zed.

A caution about OpenCode Go though, the entire company seems to be run by AI so there's lot of billing related issues with zero support. I subscribe new every month as I lost money due to double payment with automatic subscription.

For non coding related tasks I use local models.

P.S. If anyone is interested to read more about my setup, let me know I'll publish a blog post.

I agree completely. I think local AI is best limited to purpose built SLMs; all this craze around running quantized coding LLMs has taken the attention off SLMs.