Open Source AI Art w/ @banodoco | 👫🐕 w/ @hannahsubmarine

Hallihallo / Howiya
Frontier models demos used to make me sad because it was cool stuff that open/local models can’t do But now they make me happy because I see them as previews for stuff open models will be able to do in 6-9 months w/ far more freedom + at much lower cost 🤠
5
29
893
The real singularity is the joining of mankind w/ the universe Our entire history is us creating knowledge that in turn creates tools that allow us to imbue the elements of the universe w/ a purpose that no individual human decided
One way to think of AI is as something magical and unprecedented But in a time of relative hysteria, it’s also worth remembering that its also just a continuation of centuries-old trends
1
5
579
(yes, you can have borderline spiritual beliefs re: AI w/o believing that it’s a magical special technology w/ no precedent!)
1
126
In fact, I would say this viewpoint is even more magical than the narrower idea of the singularity
106
The real singularity is also the joining all of mankind together with the universe It’s also been happening for a very long time “What hath God Wrought?” - the text of the first telegraph
One way to think of AI is as something magical and unprecedented But in a time of relative hysteria, it’s also worth remembering that its also just a continuation of centuries-old trends
386
One way to think of AI is as something magical and unprecedented But in a time of relative hysteria, it’s also worth remembering that its also just a continuation of centuries-old trends
Made with AI
1
10
1,276
POM retweeted
I retrained all the Minimax H3 training adapters on generic data and they are yielding far superior results, even without contrastive guidance. These are now defaults in AI Toolkit. I also added Fast H3 with a training adapter so you can now train on it directly.
10
23
299
11,163
I’m a small model maximalist so it brings me no pleasure to say this but GPT-6-Luna is a bad release, it feels considerably worse than 5.6 Luna across the board (which is the first time in over a year that an OpenAI release hasn’t been clearly better than what it replaced)
7
63
4,059
I’ve released an experimental MiniMax H3 VFX Edit LoRA for ComfyUI. It uses your source video as a frame-aligned guide, aiming to apply only the requested change while preserving motion, timing, framing and the rest of the shot. Download: huggingface.co/Alissonerdx/M… civitai.red/models/2954203/v… #ComfyUI #MiniMaxH3 #AIvideo #VFX
8
7
51
4,662
POM retweeted
I know a lot of people are not super concerned about the licenses, and I get that. Most people have never been in a deposition being grilled for hours by a team of lawyers going over what the word "is" means in a document. I have. Most people have never been sued in a contract dispute. I have. Most people have not had to pay lawyers hundreds of thousands of dollars to defend them in court on a frivolous lawsuit that you will win, but had to fight anyway, and the legal fees alone bankrupted your company. I have. More than once. I have never lost a court case, but I have been invited to them many times. And when you are invited to one, you want to be 100% sure and confident that you are in the right. Trust me, the licenses DO matter.
28
29
343
13,042
Tried this with some of my own keyframes - "O HI"
Wow: simple painted keyframes to generate fluid sim / VFX elements This is built with an open-source IC-LoRA that uses @ltx_io - renders in @ComfyUI - made by Adrien Toupet 🤩
5
26
413
59,755
Working on a zero-shot ‘audio reference to song’ module for YuE2. provide a song for reference then pick the key, tempo and lyrics for the new song that will use the singer and band from your reference. Here’s a new video from Nora while it’s training! piped.video/20A_JM8BJnY
5
4
46
2,120
working on the Minimax H3 version of my crossview-warp lora. Based on my first tests it is quite promising. (for those who don't know the lora: it can change the camera movement on your existing video clip)
11
16
181
27,995
Every day, I come across new well-funded efforts to stoke fear of AI Doomers are giving people a religious movement that prays on their deepest fears - not just death but poverty/obsolesce - while the other side only offers just pragmatic arguments
9
1
24
2,756
I really dislike MCPs They feel so old-fashioned, like from an era where people weren’t constantly reshaping every facet of the tools they work with
13
2
38
2,816
I can 100% see why they appeal to trad programmers though and to relatively new agent-powered builders though
2
3
315
The "hum-to-song" module training is complete. I'm open-sourcing it within the hour. Follow to make sure you don't miss the update. Special thanks to my mates at fixedseed.com for sponsoring the compute. #YuE2 #opensource
18
52
411
40,755
Yann LeCun is essentially right about the most important problem of our time - that AI safety will be an incrementally-tackled engineering problem For starters, every incident so far and even of the most imaginative LW fanfics could be prevented w/ basic monitoring
It seems to me that before "urgently figuring out how to control AI systems much smarter than us" we need to have the beginning of a hint of a design for a system smarter than a house cat. Such a sense of urgency reveals an extremely distorted view of reality. No wonder the more based members of the organization seeked to marginalize the superalignment group. It's as if someone had said in 1925 "we urgently need to figure out how to control aircrafts that can transport hundreds of passengers at near the speed of the sound over the oceans." It would have been difficult to make long-haul passenger jets safe before the turbojet was invented and before any aircraft had crossed the atlantic non-stop. Yet, we can now fly halfway around the world on twin-engine jets in complete safety. It didn't require some sort of magical recipe for safety. It took decades of careful engineering and iterative refinements. The process will be similar for intelligent systems. It will take years for them to get as smart as cats, and more years to get as smart as humans, let alone smarter (don't confuse the superhuman knowledge accumulation and retrieval abilities of current LLMs with actual intelligence). It will take years for them to be deployed and fine-tuned for efficiency and safety as they are made smarter and smarter.
1
6
1,144
It seems like everyone who hates AI is actually just at the beginning of the journey towards loving open source AI 🤗
Ok, fine, programming is a little bit solved. My only regret is the whole associated mega corporation owned AI thing and the descent into dystopia.
3
1
11
1,270
You’re almost there @notch, just keep going!
1
2
231