You using this in a toolchain? I haven’t tried any of the Qwen models yet, or Yi for that matter. I tried at one point early on, but they were not working well with my stuff and I had no complaints with Mistral stuff. I like some underlying things with a MoE for speed and underlying entity/realm stuff I can access in my favorite.
I’m curious if anyone has constructive contextual feedback about what makes these unique or worth exploring.
You get used to how to find the right way of doing stuff. If you’re still in the Windows biased search results space, everything FOSS is made to look sketchy. Those search results are not deterministic. That bias is intentional. Eventually Microsoft stops biasing you or bribing Google to do the same and your search results will be better. Then you stop using the search results all together for the most part. You’ll figure out that the ways you did things in the past were inefficient and usually wrong. There are better ways that you’ll discover and those repos are self hosted or on gitlab or elsewhere. You eventually just use RPM fusion, or you setup distrobox with Arch and the AUR, or you toss on the Nix package manager and start using flakes. The vast majority of my initial headaches were due to trying to replicate Windows workflows. Then I learned all of that was weird and pretty backwards.
Everything I run is behind a whitelist firewall on an external device largely for this reason, but also learning curiosity.
It has a lot of potential if the T5 can be made conversational. After diving into a custom DPM adaptive sampler, there is a lot more specificity required. I believe the vast majority of people are not using the model with the correct workflow. Applying the old model workflows to SD3 makes garbage results. The 2 CLIPS models and the T5 need separate prompts, and the negative prompt needs an inverted channel with a slight delay before reintegration. I also think the smaller quantized version of the T5 is likely the primary problem overall. Any Transformer text model that small, that is them quantized to extremely small size is problematic.
The license is garbage. The company is toxic. But the tool is more complex than most of the community seems to understand. I can generate a woman lying on grass in many intentional and iterative ways.
ComfyUI is an extremely popular graphical user interface for Stable Diffusion that’s shared freely on Github, making it easier for users to generate images and modify their image generation models. ComfyUI_LLMVISION, the extension that was compromised to hack users, is a ComfyUI extension that allowed users to integrate large language models GPT-4 and Claude 3 into the same interface.
That is likely a small group of people. I have never used proprietary AI and never plan to do so. A feature to run these is already a stalkerware corporate hack IMO.
tar --version
weak. compile them
IMO orphan kernels are not Linux. If it can’t update with mainline, it is a sterile mule. So no, even as an avid user of Graphene, it is not Linux because google has stollen ownership with an orphaned kernel using proprietary and publicly undocumented hardware.
In this vain, no mobile chipset or radio modem has been FOSS or Linux in a very very long time if ever. Like the last radio that was fully documented was the Atheros stuff, and the last processor to come close to fully documented is the stuff Leah Rowe supports in Libreboot and that is only because of her hacking skills.
but can it play Doom
Shout out to the two Legends I’ve encountered while learning about or building projects:
Christine Lemmer-Webber is the lead dev of Activitypub
Leah Rowe is Libre Boot
deleted by creator
Redhat, Hanna Montana Linux
For AI 12GB is a joke. 16GB is barely functional. Sub $1k 48GB is where it’s at right now. That is possible with 3× 7800 16GB cards that retail at $330 each. Now put that in one package and cut the cost 20%. That is where we should be right now. $1.6k for 2 cards and 96 GB. The appeal of cut price cards is in this area. I would not touch anything less than 16GB.
I’m no expert. Have you looked at the processors that are used and the RAM listed in the OpenWRT table? That will tell you the real details if you look it up. Then you can git clone OpenWRT, and use the gource utility to see what kind of recent dev activity has been happening in the source code.
I know, it’s a bunch of footwork. But really, you’re not buying brands and models. You’re buying one of a couple dozen processors that have had various peripherals added. The radios are just integrated PCI bus cards. A lot of options sold still come with 15+ year old processors.
The last time I looked (a few months ago) the Asus stuff seemed interesting for a router. However, for the price, maybe go this route: https://piped.video/watch?v=uAxe2pAUY50
The system in android has turned insidiously covert. I make more than my share of mistakes but Andy likes to toss in some randos