#linux Bot Logged User list

Network: Rizon

Modes: +CNRntz

Last Seen: an hour ago

Topic: Welcome to #linux ! | Channel Rules: https://wiki.rizon.net/index.php?title=Linux | Ask your question(s) and be patient as it is different time(s) around the globe | NUchat IRC client: https://github.com/lord3nd3r/NUchat

#20

Rank

124

Users

Channel Log Archive for #linux

Hide Joins

* All times are UTC

Filtering by user: NoCode

Sunday, May 10, 2026

[01:37:14] @ NoCode I think I was for a while

[01:38:10] @ NoCode brothchild, Sorry for your loss.

[02:09:09] @ NoCode What's the cheapest, best processing computer you can us to get some LLMs working at a good rate, locally?

[02:09:45] @ NoCode I really need to figure something out and build something

[02:09:49] @ NoCode Oh yeah, for sure

[02:13:23] @ NoCode What I should be looking for is a dataset trained minimally on just 3-4 web dev languages, maybe python too. Or hell, look for a web stack LLM. Though, if I'm going to use something like that I'm want more things to build web apps

[02:13:55] @ NoCode Maybe Rust or Ruby or whatever is better for the occasion.

[02:14:30] @ NoCode nice

[02:15:53] @ NoCode glitchy, Please list a webstack LLM. Small model processing. low hardware spec

[02:16:19] @ NoCode rrrhm derp

[02:16:51] @ NoCode nice

[02:18:04] @ NoCode Yeah, I like agentic shit. Cursor has been fun but it's pricey as hell.

[02:19:32] @ NoCode nice

[02:19:56] @ NoCode Sounds pretty wild

[02:21:25] @ NoCode gonna check out qwen2.5-coder:3b

[02:21:31] @ NoCode but I doubt I can run it

[02:21:36] @ NoCode Client: HexChat 2.16.2 • OS: Debian 13.4 • CPU: Intel(R) Core(TM) i5-8250U CPU @ 1.60GHz (2.39GHz) • Memory: Physical: 7.5 GiB Total (2.7 GiB Free) Swap: 10.9 GiB Total (10.3 GiB Free) • Storage: 85.2 GB / 263.1 GB (178.0 GB Free) • VGA: Intel Corporation UHD Graphics 620 @ Intel Corporation Xeon E3-1200 v6/7th Gen Core Processor Host Bridge/DRAM Registers • Uptime: 2d 4h 27m 44s

[02:24:29] @ NoCode nice :D

[02:25:33] @ NoCode :)

[02:26:35] @ NoCode 3 years ago I had an aging build. I mean, I think it was 15 years old back then and ran well. I miss my nvidia 950 GPU

[02:26:57] @ NoCode Yeah, I definitely use a lot of context

[02:27:38] @ NoCode I'd rather explain to the LLM then figure out what it has coded, and manually fix it. I can think with that type of processing in knowing what I want and further explain it. :D

[02:28:41] @ NoCode this type of coding also would help me be a better coder, with examples.

[02:29:01] @ NoCode Yeah for sure, small/low power. This is the only computer I have

[02:30:40] @ NoCode for sure. Since Windows is telling people they need to upgrade their shit, there's a lot of used Windows machines on marketplace. I'm probably gonna go that route.

[02:31:37] @ NoCode nice

[02:32:10] @ NoCode welcome back sir

[02:36:24] @ NoCode Call me crazy, but I'm gonna make that coffee I've been talking about for 2 hours. 10:36 PM here, but I'm cold as heck. brb.

[02:39:07] @ NoCode End3r, Reverse engineer Donner's Hush 1 Pro software so it runs on Linux so I can mess around with adding more IRs and presets/updating firmware. :D

[02:40:14] @ NoCode I mean the interface works well when hooked up to USB to the computer. Real-time works well, I can record to a DAW fine. But not having a Linux app has be cooked. :D

[02:40:22] @ NoCode I told them to dev a Linux app

[02:41:42] @ NoCode Yeah it's fun, and was cheap

[02:42:07] @ NoCode I'm gonna buy one of their hot rodded strat style guitars

[02:42:21] @ NoCode I think I paid $350 CAD

[02:43:59] @ NoCode That's how good things start.

[02:45:02] @ NoCode nice

[02:45:33] @ NoCode :)

[02:46:09] @ NoCode ha

[02:49:47] @ NoCode those were the good days

[02:50:09] @ NoCode I remember loading 10s of thousands of tunes in Wnamp and it handled it very well.

[04:26:55] @ NoCode Have you tried, https://github.com/simonw/llm -- it looks liike a stripped down LLM management tool for the term.

[04:28:54] @ NoCode I like IDE or terminal variants because I self contain

[04:29:08] @ NoCode can*

[04:31:56] @ NoCode sentient, I think I'll try that

[04:38:53] @ NoCode error: invalid-egg-fragment with the llama option

[04:40:45] @ NoCode .msg sentient Hey man, so yeah I followed the installation option for llama, and got "error: invalid-egg-fragment" -- unless i need ollama installed first.

[04:40:50] @ NoCode oops. :D

[05:05:07] @ NoCode sentient, Got it, thx

[05:17:50] @ NoCode sentient, Ideally just looking for smaller ones to run since my spec of a computer is low

[05:20:23] @ NoCode I am all accepting monetary donations at your disposal to upgrade. :D

[06:11:25] @ NoCode sentient, Falcon-H1-0.5B-Instruct-heretic.i1-Q6_K.gguf -- Wow, this is interesting

[06:11:38] @ NoCode Best model I've used/self-hosted so far!

[06:16:57] @ NoCode I'm not too skilled at anything mind you. :D

[06:17:11] @ NoCode I'm just a luser at the moment.

[06:31:15] @ NoCode damn, this is cool. I just trained the model on a topic, and asked it to reflect on what it was trained on. And with what sources. It's interesting to see this work so well

[06:31:40] @ NoCode That's with memory:yes. Logged, etc..

[06:32:07] @ NoCode so it reflected back on the trained data, but also suggested it was trained on public record.

[06:32:29] @ NoCode so cool

[06:34:11] @ NoCode there's absolutely no struggle with this LLM on this shit-top of a laptop.

[06:34:25] @ NoCode This totally changes things

[06:51:01] @ NoCode I told it to look on Wikipedia and Amademic sources excluding JSTOR

[06:52:41] @ NoCode I asked it on the history of the bicycle, and the top 5 cyclists. Where they were from, their inspirations, and what brand they used. Or something like that.

[06:52:46] @ NoCode I worded it different.

[06:52:56] @ NoCode I removed what I said

[06:52:59] @ NoCode ;D

[06:53:15] @ NoCode It used 3 sources including Wikipedia

[06:54:10] @ NoCode When i didn't mention JSTOR, it pulled up 3 sources from JSTOR

[06:54:25] @ NoCode Also, fuck JSTOR. They can burn in hell.

[06:54:34] @ NoCode how big is it?

[06:57:24] @ NoCode it says, "search" in Meltdown"

[06:57:34] @ NoCode it says, "search" in Meltdown.*

[06:57:41] @ NoCode Click yes

[06:58:40] @ NoCode this is as fast as Cursor, and better IMO

[06:58:59] @ NoCode Them mofos aren't getting my data.

[07:08:31] @ NoCode would text to image work in meltdown?

[07:09:12] @ NoCode cool

[07:09:49] @ NoCode Thanks for suggesting this interface. It feels way less clunky than ollama and that LM app

[07:09:54] @ NoCode there's no clunk at all.

[07:10:04] @ NoCode Straight to business

[07:10:28] @ NoCode When I seen tkinter, I was "....f, yeah!"

[07:11:40] @ NoCode It reminded me of trying to learn python, but realized back then it was just a library and I kinda stopped trying to learn it. Getting knee deep into a internet browser :D

[07:12:14] @ NoCode from zero coding experience

[07:12:21] @ NoCode probably like 20 years ago

[08:06:48] @ NoCode I wonder if Rpi3's would be good for small LLMs?

[08:07:55] @ NoCode It'd be cool to use 2 rPis. One for orchestrating/thinking/planning, while the other follows the instructions. So they offload each other

[08:10:39] @ NoCode nice

[08:35:24] @ NoCode I used to use IRSSI back in the day. :)

[08:36:07] @ NoCode couldn't figure out BitchX and it felt like I was gimped. So I went onto IRSSI

[08:36:13] @ NoCode and man, that was cool

[16:19:06] @ NoCode Interesting, I found an low profile agentic IBM coder. I mean, if the names checks out, it should be goo right?? I asked it to list optimizations to the code I've been vibe coding for a while, and it listed a bunch of optimizations. So I think I'll run the changes through Cursor.

[16:19:26] @ NoCode IBM-Agentic-Nvidia-Q4_K_M.gguf

[16:24:37] @ NoCode At this point it told me to start using a framework

[16:30:08] @ NoCode What can I use as a IDE with a locally hosted LLM? Is there a way to use OpenCode or VSCode?

[17:28:08] @ NoCode weasel, Who da fuck is this ferret.

[17:28:15] @ NoCode Number tree fiddy

[17:29:25] @ NoCode Just made some Seitan, it's steaming now. Coffee #1 of the day

[17:30:20] @ NoCode cool

[17:30:34] @ NoCode vscode + continue.dev

[17:30:45] @ NoCode Shit's about to get real

[17:31:34] @ NoCode I have no idea why I didn't ask glitchy.

[17:31:37] @ NoCode brain fart

[17:31:50] @ NoCode glitchy, tell me to stop farting brains

[17:32:16] @ NoCode glitchy, Thanks to your master.

[17:32:44] @ NoCode k

[17:58:31] @ NoCode seitan is done and it tastes good

[18:04:08] @ NoCode Not really, it fits the protein that is needed without the cost. Yup, add nutritional yeast or pea protein for essential aminos. I added pea protein.

[18:04:42] @ NoCode Why are you turning it in an argument?

[18:05:26] @ NoCode yep

[18:05:36] @ NoCode the focus isn't the gluten.

[18:06:39] @ NoCode woodwose, I buy it pre-separated out of convenience.

[18:08:44] @ NoCode I can make TVP taste like meat. But I don't want the soy like that. Once in a while is alright

[18:09:00] @ NoCode I can make some killer spagbol with TVP

[18:09:33] @ NoCode Anyway

[18:12:01] @ NoCode yep

[18:14:12] @ NoCode Out of arrogance, you just threw what I said out the window.

[18:14:23] @ NoCode To win what?

[18:15:10] @ NoCode glitchy, Give woodwose a reward.

IRC Log Archive

This is a searchable archive of IRC conversations from #linux on Rizon. Use the search box above to find discussions on specific topics. All conversations are preserved to help the community learn from past discussions.