Gracana

15 ms·

211.
by Gracana 1y ago
You should stick with extensions that have lots of stars, that way you know they're trustworthy and secure.
212.
by Gracana 1y ago
https://huggingface.co/LoneStriker for exl2 quants.
213.
by Gracana 1y ago
I see your frustration that these people exist who don’t share your values, but their comments already get downvoted. Take the win and move on.
214.
by Gracana 1y ago
This Old Tony (a hobby machinist / welder on youtube) has made all of his videos in that format. It works very well! https://www.youtube.com/c/thisoldtony/videos
215.
by Gracana 2y ago
Awesome. It would be great to see a nice gtk-based open source competitor to lm-studio and the like.
216.
by Gracana 2y ago
You can tell it to try again. It took me a couple rounds with the tool before I noticed that your conversation after the initial research isn't limited to just chatting: if you select the "deep research" button on your mess
217.
by Gracana 2y ago
This looks quite nice. I would like to see the system prompt and inference parameters exposed in the UI, because those are things I'm used to fiddling with in other UIs. Is that something that the llm library supports?
218.
by Gracana 2y ago
Nope, sounds neat though. There's so much to keep up with in this space.
219.
by Gracana 2y ago
You can probably find some long-form tuned models on HF. I've had decent results with QwQ-32B (which I can run on my desktop) and Mistral Large (which I have to run on my server). Generating and refining an outline before writing the w
220.
by Gracana 2y ago
That is way cooler than I expected.
221.
by Gracana 2y ago
Nice. Among the things other folks have mentioned, you'll probably want to figure out a different cooling fan. My recollection is that the stock fan is annoyingly loud. Oh hey, I figured I would take a look on thingiverse for a solutio
222.
by Gracana 2y ago
Beige tower, desktop, or the blue and white tower? I always wanted a B&W G3, my friend('s parents) had one and it was a beautiful machine. I had the desktop "beige G3" and I ran it for a long time. Added a 433MHz G4 accel
223.
by Gracana 2y ago
People thought that the speed itself was dangerous, that the wind and vibration and landscape screaming by at 25mph would cause physical and mental harm.
224.
by Gracana 2y ago
These are important things to look at because they're where LLMs typically have trouble. People complain that their stories are too short, that they use stock phrases and purple prose, that they start to repeat themselves at a certain
225.
by Gracana 2y ago
I don't know what your question means. OP said "why should I be bothered to read what nobody could be bothered to write," which I think is silly, because I enjoy all kinds of things that nobody bothered to create. Things that
226.
by Gracana 2y ago
Why limit yourself only to experiences curated by humans? You wouldn't do that outside of reading, it'd be an impossible limitation to impose. The only reason reading has been that way is because only humans could write. Now machi
227.
by Gracana 2y ago
That's my experience as well. If you feed your summary + outline + guidance and prompt a one-shot output, it'll rush through it. If you prompt it for longer length, it'll extend it for little benefit. To get good output, you
228.
by Gracana 2y ago
Yup, that's the problem I run into. You give it some lore to draw on or describe a character or give them some knowledge, and it'll just blurt it out when it finds a place for it. It takes a lot of prompting to get it to stop, and
229.
by Gracana 2y ago
I've been using QwQ-32B a lot recently and while I quite like it (especially given its size), I noticed it will often misinterpret the system prompt as something I (the user) said, revealing secrets or details that only the agent is su
230.
by Gracana 2y ago
You can get Mini-ITX server motherboards with IPMI and ECC memory if you want something robust and remotely-manageable while staying small and low power. I have a SuperMicro A2SDi-4C-HLN4F, which has a 16W intel atom C3558 and is quite old
231.
by Gracana 2y ago
I am immediately reminded of this: https://www.youtube.com/watch?v=Hv6RbEOlqRo
232.
by Gracana 2y ago
Forrest Mims books were great for me as a kid who wanted to learn about electronics. I'd recommend Getting Started In Electronics, which you can still buy new. There's a copy on archive.org, but I'd recommend getting a copy..
233.
by Gracana 2y ago
Based on the chip count in the promo picture it looked like it was going to be a 256 bit interface, but it'd be unfair to say that it was known at that point. I certainly wish it was better.
234.
by Gracana 2y ago
That's what everybody was predicting based on the memory technology and the pictures. I'm just glad to hear about DGX Station, though I doubt I'll be able to afford one.
235.
by Gracana 2y ago
A lot of games store save files in appdata. That’s the only thing I (not op) can think of, and one I regularly find annoying (if only slightly).
236.
by Gracana 2y ago
It does have an edge on being able to run large MoE models.
237.
by Gracana 2y ago
Absolutely. The thinking models really hurt. I was happy with anything that ran at least as fast as I could read, then "thinking" became a thing and now I need it to run ten times faster. I guess code is tough too. If I'm tal
238.
by Gracana 2y ago
Despite the hype, the 512GB Mac is not really a good buy for LLMs. The ability to run a giant model on it is a novelty that will wear off quickly... it's just too slow to run them at that size, and in practice it has the same sweet spo
239.
by Gracana 2y ago
Looks like it wants https://github.com/marp-team/marp-cli
240.
by Gracana 2y ago
That’s wild, somehow I hadn’t seen the B200 specs before now. I wish I could have even a fraction of that!

More ›