Apple has some interesting text-to-image and image-to-image models they've been working on. Rumor is we'll be getting an overhauled Siri with an Apple-built LLM announcement at WWDC. But I don't have concrete evidence of that yet.
AI
-

Text-to-Video AI Joining Generative AI Revolution
By
–
So excited to see text to video joining the party. You can learn more about this specific task here including with some open-source models and datasets: https://
huggingface.co/tasks/text-to-
video
…. AI is going to transform every single modality, domain and tasks of technology building! -
Exciting technological improvements coming for longer video capabilities
By
–
So I saw a few things that led me to that conclusion. Not one single thing… There's more exciting stuff coming that will allow us to go even longer with videos… 🙂
-
OpenAI ChatGPT Updates: This Week’s Major AI Announcements
By
–
I'm planning on making another "This Week in AI News" style video for tomorrow and it's probably going to have another "Insane Week" kind of title that so many people tell me they hate. LOL But here's what's happened in the past 72(ish)-hours: – OpenAI announced that ChatGPT
-
Major AI Video Breakthrough Moment Comparable to Midjourney Evolution
By
–
This is the moment for AI video that we experienced when we saw the leap from MJ3 to MJ4.
-

Rethinking Machine Unlearning for Large Language Models
By
–
Rethinking Machine Unlearning for Large Language Models Liu et al.: https://
arxiv.org/abs/2402.08787 #ArtificialIntelligence #DeepLearning #LargeLanguageModels -
VQVAE Stack Architecture Differs Fundamentally from Transformers
By
–
also a “VQVAE stack” is not a transformer! it would not look like this haha
-
Fine-tuning GPT2 with additional layers avoids pretraining scratch
By
–
he’s pointing out we can just add the two layers I drew here and train those starting from pretrained GPT2 or something; don’t have to pretrain from scratch
-
Discrete Tokens Recover AI Model Performance Effectively
By
–
if you use enough discrete tokens you could def recover performance eventually, unless you’re doin somethin wrong
-
Multi-token sparsity in discrete token activations
By
–
yeah it’s similar— I think you could call it multi-token sparsity or something like that? I mean obviously the activations have to be sparse in this case but we can use as many discrete tokens as we want (unlike in the projects you’re mentioning)