computer science
Zhejiang
Joined April 2019
- Tweets17
- Following798
- Followers12
- Likes14
canopus retweeted
There’s a single formula that makes all of your diffusion models possible: Tweedie's
Say 𝐱 is a noisy version of 𝐮 with 𝐞 ∼ 𝒩(𝟎, σ² 𝐈)
𝐱 = 𝐮 + 𝐞
MMSE estimate of 𝐮 is 𝔼[𝐮 | 𝐱] and would seem to require P(𝐮|𝐱). Yet Tweedie says P(𝐱) is all you need
1/3
canopus retweeted
0/n #Midjourney 写prompt是一件很痛苦的事情，感觉像AI在训练我们而不是我们在训练AI。但我今天找到了几个宝藏prompt网站，里面有很多已经训练好的风格指令，帮我们一键生成想要的风格。
canopus retweeted
"You are GPT‑3, and you can't do math": Prompting GPT‑3 via zero-shot instruction to answer calculation/math questions by consulting a Python REPL.
canopus retweeted
"You are GPT-3", revised: A long-form GPT-3 prompt for assisted question-answering with accurate arithmetic, string operations, and Wikipedia lookup. Generated IPython commands (in green) are pasted into IPython and output is pasted back into the prompt (no green).
canopus retweeted
🌸@BigscienceW BLOOM's intermediate checkpoints have already shown some very cool capabilities!
What's great about BLOOM is that you can ask it to generate the rest of a text - and this even if it is not yet fully trained yet! 👶
🧵 A thread with some examples
canopus retweeted
For 111 days, we've enjoyed world-class hardware stability and throughput thanks to the hard work of our friends at @Genci_fr, @INS2I_CNRS, Megatron & DeepSpeed. Having reached our objective earlier than expected, we'll keep training for a few more days. Stay tuned, more soon ;)
canopus retweeted
Social distancing is needed to keep us safe. So @LandingAI created this AI tool to help our customers in essential industries maintain #SocialDistancing in the workplace. We hope many people can benefit from these ideas. bit.ly/3ew1u0C
canopus retweeted
Turn #Python Scripts into Beautiful ML Tools -
with Streamlit, an app framework built for #MachineLearning engineers buff.ly/2nmwYk4
canopus retweeted
An interesting trend from this year's CVPR are the numerous new papers on self-supervised learning. Andrew Zisserman gave a nice tutorial: project.inria.fr/paiss/files… although, there is a lot more geometry-related work as well (e.g. self-supervised depth & friends).
canopus retweeted
Nice collection of PyTorch (and some TF) Jupyter notebooks for everything deep learning by Sabastian Raschka.
github.com/rasbt/deeplearnin… github.com/rasbt/deeplearnin…
canopus retweeted
Tom Goldstein (Maryland U) on "A theoretical look at adversarial examples: a perspective from high-dimensional geometry" asking the fundamental question whether Adversarial examples are inevitable? And the answer is ... yes!
canopus retweeted
One thing you can do to make the model load over the wire faster is pass --quantization_bytes=1 to the converter (using 8 bit quantization as network compression). You'll lose a little accuracy but loading will be much faster!
