8/ Teaching Arithmetics to Small Transformers – trains small transformers on chain-of-thought data to improve accuracy and convergence speed and highlights the importance of high-quality instructive data for rapidly eliciting arithmetic capabilities.
INNOVATION
-
AnimateDiff: Motion Module for Personalized Animated Image Generation
By
–
9/ AnimateDiff – appends a motion modeling module to a frozen text-to-image model, which is then trained and used to animate existing personalized models to produce diverse and personalized animated images. https://t.co/L4VOSdideP
— DAIR.AI (@dair_ai) 16 juillet 20239/ AnimateDiff – appends a motion modeling module to a frozen text-to-image model, which is then trained and used to animate existing personalized models to produce diverse and personalized animated images.
-

LLMs as General Pattern Machines: In-Context Learning and Robotics
By
–
6/ LLMs as General Pattern Machines – shows that without any additional training, LLMs are general sequence modelers, driven by in-context learning; applies 0-shot capabilities to robotics and transfer the pattern among words to actions.
-

HyperDreamBooth: 25x Faster Personalized Text-to-Image Model
By
–
7/ HyperDreamBooth – introduces a smaller, faster, and more efficient version of Dreambooth; enables personalization of text-to-image diffusion model using a single input image, 25x faster than Dreambooth.
-

NaViT: Flexible Vision Transformer for Any Resolution
By
–
5/ Patch n’ Pack: NaViT – a vision transformer for any aspect ratio and resolution through sequence packing; enables flexible model usage, improved training efficiency, and transfers to tasks involving image and video classification among others.
-

LongLLaMA Extends Context Length via Contrastive Training
By
–
4/ LongLLaMA – employs a contrastive training process to enhance the structure of the (key, value) space to extend context length; presents a fine-tuned model that lengthens context and demonstrates improvements in long context tasks.
-
Can Generative AI Legally Hold Patent Rights?
By
–
The debate on AI, in this case Generative AI, being capable of invention in legal terms in particular the stents continues. I remain skeptical that the AI itself should be awarded patent rights as the AI model is not a legal person and we then get into issues about permission to
-
Synthetic surveying and personality simulation for customer development
By
–
A. Synthetic surveying / customer development is a cheaper alternative to doing the real thing, which is useful for ideation B. It makes customer development available even to people who don’t have large distribution C. The effective simulation of personality is a game changer
-
Mobile Phones: Fastest Adopted Human Technology in History?
By
–
In the 40 years since their introduction mobile phones have gone from 0 to almost universal adoption. Is it the fastest, most widely proliferating human-made technology in history? https://
statista.com/forecasts/1143
723/smartphone-users-in-the-world
… -
Intelligence Becomes Commodity: Preparing for AI Democratization
By
–
Intelligence will soon become a commodity, widely available to all. Just as smartphones are now in the hands of 5.2B people on the planet today. Face the facts, to confront what’s coming
