Peter Simon

The science is sciencing, the math is mathing, the stats are . . . stat . . . ing? 🤔📈🧑‍🔬 Not ONLY did we break the record for new monthly submissions AGAIN last month (21,076 in May!) arXiv is on the verge of hitting 2.5 million articles total. 👀👀👀 #TONIGHT

🥞 Your Daily Dose of AI: Unveiling the Magic of Stable Diffusion 3 youtu.be/ZJuWRx6VFCw #stablediffusion3 #stablediffusion #texttoimage #text2image #AIRevolutionAIefficiency #scalableAI #futureofAI #neuralnetworks #AIinnovation #machinelearning #technews

🍎 3 min Summary video of Apple's SDK's with AI capabilities! youtu.be/c7JxgbO9pEY #apple #ai #SDK #swift

😃 My new video Check it out: Your Daily Dose of AI: Top 5 Apple Intelligence Features from WWDC 2024 youtu.be/BDsjmyBdqS4

hyper-parameter tuning

3D modelling is dead 1 artist, 100x your workflow with CityBLD Here is New York City recreated in Unreal. A totally playable 1:1 representation of the city made by one person 🤯 The tools that creatives will have at their disposals are pure wild fire A thread🧵

We’re excited for even more people to be able to build with Llama 2 on watsonx.ai!

We believe an open approach is the right one for the development of today's Al models. Today, we’re releasing Llama 2, the next generation of Meta’s open source Large Language Model, available for free for research & commercial use. Details ➡️ bit.ly/3Dh9hNp

Exciting Announcement! Stability AI proudly presents Stable Doodle, our groundbreaking sketch-to-image tool! 🖌️Unleash your creativity as a simple drawing transforms into a vibrant, dynamic image. #StabilityAI Read more bit.ly/44CV0Xr

zeroscope_v2 XL, A watermark-free Modelscope-based video model capable of generating high quality video at 1024 x 576 Model on @huggingface : huggingface.co/cerspense/zero… This model was trained with offset noise using 9,923 clips and 29,769 tagged frames at 24 frames, 1024x576 resolution.
zeroscope_v2_XL is specifically designed for upscaling content made with zeroscope_v2_576w using vid2vid in the 1111 text2video extension by kabachuha. Leveraging this model as an upscaler allows for superior overall compositions at higher resolutions, permitting faster exploration in 576x320 (or 448x256) before transitioning to a high-resolution render. zeroscope_v2_XL uses 15.3gb of vram when rendering 30 frames at 1024x576

Happy to announce #SDXL 0.9 release you can try out @clipdropapp, researchers can get weights too via below link. After feedback aiming to release 1.0 CreativeML mid-July (🤞) Trying to get it to a single model vs two Great work by @robrombach & team clipdrop.co/stable-diffusi…

3D-aware Image Generation using 2D Diffusion Models proj: jeffreyxiang.github.io/ivid/ abs: arxiv.org/abs/2303.17905

GN

A Survey of Large Language Models arxiv.org/abs/2303.18223

There's a lot of folks under the misunderstanding that it's now possible to run a 30B param LLM in <6GB, based on this GitHub discussion. This is not the case. Understanding why gives us a chance to learn a lot of interesting stuff! 🧵 github.com/ggerganov/llam…

Should we train our own large language model (LLM) on domain-specific data from scratch? Researchers at Bloomberg did just that and shared a detailed technical report describing the dataset, model configuration, and training procedure. 1/9

Stability AI is excited to announce the launch of Stable Diffusion Reimagine! We invite users to experiment with images and ‘reimagine’ their designs through Stable Diffusion. More info → stability.ai/blog/stable-di…

Attention all #NLP enthusiasts! Yesterday @PyTorch 2.0 officially got released! 💥 Check out How to get started with PyTorch 2.0 and Hugging Face Transformers and learn how to get up to 2x speed improvements. 🏎 👉 philschmid.de/getting-starte… philschmid.de/getting-starte…

PD 2.0 = Pandas 2.0 Release - It have basically become a FAST BIG data framework - pandas.pydata.org/docs/dev/whats… - airbyte.com/blog/pandas-2-…

With the release of PyTorch 2.0, we're sharing the latest for Stable Diffusion on Hugging Face Diffusers! Using the optimized scaled dot-product attention in PyTorch and torch.compile(), the team achieved up to 50% speed up across different accelerators. hubs.la/Q01HdV_Z0