Thread 1/3
How would an early version of AGI influence social media in its current form?
Will enable language, intention, sentiment, meaning and emotion level control of individual users posts, interactions and content
Risks:
1. Insights from peoples opinions can be used
A general-purpose material property data extraction pipeline from large polymer corpora using natural language processing
demo: https://t.co/KNJ49CWcOQ
repo: https://t.co/pdS4Tw4ooA
model: https://t.co/R9rLOF6N05
paper: https://t.co/fBxjO0yxYn
Oldies but goldies: G. H. Golub, Christian Reinsch, Singular value decomposition and least squares solutions, 1970. The most popular algorithm to compute efficiently the SVD decomposition. https://t.co/3cNGVndV1M
All major social media, ad monetizarion corporations need to be regulated 10x more. Also people need to learn how to make morally right decisions even at the cost of appearing successful.
Yesterday I gave a lecture at @Stanford's CS25 class on Transformers!
The lecture was on how “emergent abilities” are unlocked by scaling up language models. Emergence is one of the most exciting phenomena in large LMs…
Slides: https://t.co/Bqht9OEw7m
InstructPix2Pix by Tim Brooks allows you to write natural language instructions to edit images ✏️🖼️
We are getting closer and closer to "photoshop with words"! 🎨
Play with it now on @huggingface Spaces https://t.co/GdeFogJ8ZK
@francoisfleuret Agree. Many things that are simple are totally lost when you have school boards and committees that make decisions. The problem of bureaucracy. I personally will not be using chatgpt to help me with anything.
Muse: Text-To-Image Generation via Masked Generative Transformers
Presents Muse, a text-to-image Transformer model that achieves SotA image generation perf while being far more efficient than diffusion or AR models.
proj: https://t.co/JljH4pGZX1
abs: https://t.co/52bc9HuT7o
I don’t know why but the cyber truck does not seem sustainable or eco friendly at all. I mean that was the whole point of their cars - nothing about the billionaire world makes any sense
Roll for initiative!
This year, we partnered with industry, academia, and other government agencies in a series of 10 tabletop exercises to map out how electric air taxis will fit safely into the national airspace. Read about these expert-led discussions: https://t.co/D4MlAw6jso
@VitalikButerin I think the fed and the US monetary policies overrides all cultural norms and reforms. But current monetary policies don’t account for the difference between technology driven and non technological areas of the economy. It’s most likely going to get worse
@VitalikButerin I think the fed and the US monetary policies overrides all cultural norms and reforms. But current monetary policies don’t account for the difference between technology driven and non technological areas of the economy. It’s most likely going to get worse
A second is equal to the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the unperturbed ground state of the caesium 133 atom.
Got some great feedback. To clarify: (1) DALLE-1 is a pure transformer, but *not* diffusion-based. It uses VQVAE to render the image given a seq of latent vectors. (2) DALLE-2 and SD of course still use Transformers in text encoder and CLIP. DiT replaces their U-Net renderer.
It is likely that the current manifold and geometric based approach to robotics will replace the Kalman filter EKF and similar non linear approximation methods.
Introducing RT-1, a robotic model that can execute over 700 instructions in the real world at 97% success rate!
Generalizes to new tasks✅
Robust to new environments and objects✅
Fast inference for real time control✅
Can absorb multi-robot data✅
Powers SayCan✅
🧵👇