keriati1

So a few months ago for various reasons I started to use ubuntu/gnome again (after 15 years on MacBooks / OS X / MacOS) on my home PC. I have to say, in my case, command line llm made the desktop so much more enjoyable.

Sadly still a lot of things did not work correctly out of the box, and everything was kinda fixable, but would have needed some bigger time investment from my side.

Instead just telling llm cli that something does not work correctly, and waiting a few minutes until it fixes it is just so much better for me.

Do I want LLMs all over the place in my desktop? Probably not. But having an LLM fixing all kind of broken or miss-configured software is just so convenient.

I wanted to get a JetKVM, however ended up getting kvm extension module for my uConsole instead. In the end it is perfect for my small collection of home servers. I only use it a few times a year, but it's always fun.

Yes I tried it for a while, it can be nice.

A few years back I started to try to use my computer by not touching the mouse. Ended up writing a bunch of Tampermonkey scripts for my most used web pages to add fast keyboard navigation shortcuts.

Also a lot of webpages have it already built in, by pressing "?" they show a nice overlay. For example github and gmail have it.

For github I still ended up adding a quite a few more shortcuts.

Can confirm, this is the situation.

Getting messages at 7am that whole CI broke over night, tests and builds are dead, developers can't run install anymore. And as in any enterprise, everyone looks at the proxy, what could be wrong with it that it blocks the audit endpoint.

I work for an old enterprise, so far rather conservative with LLM/AI usage. However the copilot cli adoption in the last 2 weeks is spreading light wild fire. Codex 5.3, a good instructions file and it works. Features are getting done and delivered in days, proper test coverage is done, proper documentation in place. Onboarding to it is also very fast.

I think this might be very misleading. I own (in Europe) my 3rd VW.

The reality is, that this check is done in the VW car service, and they do a pre-check. Then they call me on the phone and tell me what they need to fix and how much it will be. I always tell them OK, and they just fix everything before the real check is done. And this is exactly how it always works for all the people I know.

Also, as far as I know, Tesla owners just take the car to a "random" check point and are probably surprised why it didn't pass.

Today github labels, tomorrow paperclips?

I’m not sure why, but the first thing I did was check if HTTP status code 418 was listed.

I can also recommend rather to use the stacked PR approach. We have it since years, PR review "issues" are not a thing for us.

I still encourage do to a lot of small commits with good commit messages, but don't submit more then 2-3 or 4 commits in a single PR...

What model size is used here? How much memory does the GPU have?

We had some as build agent around already. We don't plan to do any fine tuning or training, so we did not explore this at all. However I don't think it is a viable option.

We run coding assistance models on MacBook Pros locally, so here is my experience: On hardware side I recommend Apple M1 / M2 / M3 with at least 400Gb/s memory bandwidth. For local coding assistance this is perfect for 7B or 33B models.

We also run a Mac Studio with a bigger model (70b), M2 ultra and 192GB ram, as a chat server. It's pretty fast. Here we use Open WebUI as interface.

Software wise Ollama is OK as most IDE plugins can work with it now. I personally don't like the go code they have. Also some key features are missing from it that I would need and those are just never getting done, even as multiple people submitted PRs for some.

LM Studio is better overall, both as server or as chat interface.

I can also recommend CodeGPT plugin for JetBrains products and Continue plugin for VSCode.

As a chat server UI as I mentioned Open WebUI works great, I use it with together ai too as backend.

I think it is even easier right now for companies to self host an inference server with basic rag support:

- get a Mac Mini or Mac Studio - just run ollama serve, - run ollama web-ui in docker - add some coding assitant model from ollamahub with the web-ui - upload your documents in the web-ui

No code needed, you have your self hosted LLM with basic RAG giving you answers with your documents in context. For us the deepseek coder 33b model is fast enough on a Mac Studio with 64gb ram and can give pretty good suggestions based on our internal coding documentation.