gunalx

13 ms·

271.
by gunalx 2y ago
I found it to be a pretty OK first draft character creator, scene described and occasional lore/world builder. Not really anything more. Kinda a DM assistant more than a DM.
272.
by gunalx 2y ago
Cool real life use Case. Don't think lmms usually get applied reasonably where they should be and I am glad that a generic knn model also was used to simplify costs and also just more suitable.
273.
by gunalx 2y ago
I agree. The 2gb skew is borderline useless in most applications you wouldnt just use a pico these days. (I would even line a 32gb skew but realise that is unreasonable for the current soc)
274.
by gunalx 2y ago
Arm nixpkgs build server is one reason i can think of. (But then 16gb is still almost to little)
275.
by gunalx 2y ago
Simple and somewhat decent intro to gpt in general. But I still feel this was somewhat repetitive.
276.
by gunalx 2y ago
Why lock down to a nocode tool instead of just hacking together a simple html5 page? Ai assistants have speed up the prototype workflow for doing less code/design.
277.
by gunalx 2y ago
Mixing filaments has been tested and wastes an enormous amount of filament, and has accuracy issues with color and deposition (like almost all regular multicolor getting clean seperstions or transitions is hard)
278.
by gunalx 2y ago
The main reason not to buy bambu.
279.
by gunalx 2y ago
https://archive.is/pvW9N
280.
by gunalx 2y ago
One of the greater hacker conferences in the year. (I am a huge fan of CCC and all events they do)
281.
by gunalx 2y ago
Why is this generic news article on hacker news. The guidelines clearly states anything you would probably find in regular National news probably shouldn't be on hacker news. I want to get a break from the regular newscycle when goin
282.
by gunalx 2y ago
I asked it "show me cityname" and it did not find it. First asking for country did not help at all either.
283.
by gunalx 2y ago
Don't buy. Their standalone hardware unit has lots of hardware ground issues.
284.
by gunalx 2y ago
Seems like it. They even have example training scripts available. https://github.com/AnswerDotAI/ModernBERT/blob/main/examples... Check out their documentation page linked on the bottom of the article.
285.
by gunalx 2y ago
Its not really that bad. Being petty for pocket change and complaining on hn is lowlife. Wealth tax aint great if you want to keep the money stale, but that just means you should use it to compund itself.
286.
by gunalx 2y ago
Worked pretty terrible at trying to model the nix snowflake. Guess its more trained on natural and biologic structures and textrues, rather than more structural or symetric data.
287.
by gunalx 2y ago
Thanks for the info. I just assumed LLM was a en var manager for openai stuff. But makes sense.
288.
by gunalx 2y ago
Does it accept custom openai endpoints and model names? Would be cool to be able to hook this up with ollama or something local.
289.
by gunalx 2y ago
Switched to arch and nixos, diffrent machines and diffrent usecases. Arch installer is really nice now, even if you should have done it manually at least once in your life.
290.
by gunalx 2y ago
I think berts are still used for stuff like embeddings (at least the best generic embedding model in my language is bert based), when you want to get a model that has basic semantics but needs to be fast.
291.
by gunalx 2y ago
Is it just me or does the author swap descriptions of the instruction finetuned and the base gpt-3.5-turbo? It seemed like the best model was labeled instruct, but the text saying instruct did worse?
292.
by gunalx 2y ago
Manjaro has been dead to me for a long time (for different reasons). But this sets it in stone. I used to like Manjaro as a simple stable arch Alternative. Telemetry should always be opt in.
293.
by gunalx 2y ago
Really cool tech demo. What for the most part impressed me is the inference speed. But I don't really see any use for this unless a way to store worldstate to avoid the issue of it forgetting what it just said.
294.
by gunalx 2y ago
Yes. Especially if you work in a not well supported language and/or have specific datapairs you want to match that might be out of ordinary text. Training your own fine tune takes a really short time and GPU resources, and you can eas
295.
by gunalx 2y ago
I do get the point of using form validation clientside to ensure a better ui. But dont remenber to also verify serverside. Anything clientside can have been fumbeled with. (Also kinda anoying to have to duplicate this tho)
296.
by gunalx 2y ago
Why the .exe name when it seems to be intended as a multiplatform support with macOS as main?
297.
by gunalx 2y ago
Exactly what i need to be able to not depend on m0zilla
298.
by gunalx 2y ago
Not having open ish weigths is a total dealbreaker for me. The only really compelling reason behind sub 6B models is them being easy to run on even consumer hardware or on the edge.
299.
by gunalx 2y ago
Bark?
300.
by gunalx 2y ago
But why would you run nixos as a micro vm os. It is not really designed for that purpose. Nixos as a general purpose headless server OS with nix defined services, or as a desktop system with declarative and portable config is pretty neat.

More ›