Strong work once again, @gavinpurcell. You keep pushing the edge with your work. I think you are creating a new form of educational, entertaining art. This is such an ingenious way to use one's love for AI video creation and timely commentary in a way that is easy to digest. Looking forward to the next one.
We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:
Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:
Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:
Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:
Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!
In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.
@karpathy Thank you, @karpathy. This is the type of creative and imaginative thinking needed to get the most out of this technology. It's really refreshing to see how creativity in action is pushing the frontier further. Keep the insights coming. Thanks again.
@gavinpurcell Gavin, keep up the good work. These videos are truly amazing, and they offer a glimpse of what's to come. Imagine what will be possible in 3 months. Keep creating. And โAy, I see what you did there.โ
This panic is mostly a Western cultural phenomenon. People who know better need to fight against this made-up, fantastical story. The fact that this narrative has caught on and spread so rapidly is directly related to our lack of scientific grounding.
However, this is not an excuse because we have many scientifically minded individuals who know better and understand the science of what is going on. If you find it hard to communicate your knowledge, many tools or people can help you improve your messaging.
We need knowledgeable individuals to step up, demystify this moment, and call misinformation by its name, rebutting it with facts and not using this moment to score political or financial points at the cost of our ignorance in the subject matter.
Your moment is now. Society sorely needs you. The question is who will rise to the moment.
@mreflow@smoothmedia_@davidheuff Enjoy your moment; well deserved. You really are an inspiration. I've learned a lot from you over the years. So, thank you for showing up and sharing your knowledge.
@LudovicCreator Well done, it really looks great. Also, it's a good prompt.
I tried it with seedance 2.5 a few times; it got blocked with a restriction warning, but I tried it again, and it went through only at a max of 15 seconds, not at 30 seconds.
Thanks for sharing.
Thank you for consistently making the music creation on your platform enjoyable.
As a premier user, I have a request: we really need a shuffle option in the app. This has been a request for almost a year. This makes it easy to rediscover old favorites.
Thank you for working on it hopefully we'll see it sooner than later.
Today we're also opening the weights for Muse Glimmer, a great 30B parameter dense model that can run locally. Soon we'll also release the weights for Muse Spark 1.2, our latest foundation model. Meta is a strong supporter of open source and I'm proud of these releases. Congrats to @alexandr_wang and the MSL team for all your great work on these models.
@liandeli2 Wow, Just Wow.
These video models are really amazing in the level of realism they create.
The multiple features, effects, and choreography all sync to the beat, with such clean cuts, which is astonishing.
Good job, Minimax.
@dreamina_ai This is truly amazing what you have created, and to think it only gets better from here. Thank you for making these tools available. Veo, Kling, "Sora," you guys are up next.