The George Li @thegeorgeyli

this is the most interesting year in human history, except for all future years

"A girl in a pink gown with jewelry against the backdrop of a mountain range during winter" By Gencraft: the world's fastest (and free!) AI image and video generator

REAL-TIME object detection WITHOUT TRAINING YOLO-World is a new SOTA open-vocabulary object detector that outperforms previous models in terms of both accuracy and speed. 35.4 AP with 52.0 FPS on V100. ↓ read more

I used to find writing CUDA code rather terrifying. But then I discovered a couple of tricks that actually make it quite accessible. In this video I introduce CUDA in a way that will be accessible to Python programmers, and I even show how to do it all in @GoogleColab!

“Are Transformers Effective for Time Series Forecasting?” represents a pivotal paper, decisively highlighting the shortcomings and deficiencies in research surrounding the use of transformers for #timeseries #forecasting. This paper effectively exposes the deceptive practices…

Today we’re releasing Code Llama 70B: a new, more performant version of our LLM for code generation — available under the same license as previous Code Llama models. Download the models ➡️ bit.ly/3Oil6bQ • CodeLlama-70B • CodeLlama-70B-Python • CodeLlama-70B-Instruct

RoMa: an easy-to-to-use, stable and efficient library to deal with rotations and spatial transformations in PyTorch. Read all about this PyTorch Ecosystem Tool in our latest Medium post ⚡hubs.la/Q02hFsVf0

Friends don't let friends make bad charts! Chenxin Li, pulled together a lot of great advice for data visualization, with clear "do this, not that" examples for each item. Here are a few of my favorites, see the link below for more.

keyboard_arrow_left Previous keyboard_arrow_right Next

Stanford CS25 - Transformers United Great set of lectures on advanced applications and topics related to Transformers. My favorite lectures are the ones on common sense reasoning and generalist agents in open-ended worlds. Some of the lectures are a bit old now but the…

Did you know that with 2 Python libraries, 6 lines of code and around 15 seconds, you can load satellite data from anywhere in the world? This is so much easier than it used to be!

nye in Seoul 🔔

The electronic version of the new fourth edition of my book Linear Algebra Done Right has now been downloaded over 80 thousand times in the two months since it was released. The electronic version of this Open Access book is freely available at linear.axler.net. The print…

Here are 300 hours of curated courses focused on Machine Learning Engineering. There are 15 courses. From beginner to advanced. From Google. For free. Some of the topics they cover: • Fundamentals of Machine Learning • Feature Engineering • Production Machine Learning…

Phi-2 is a damn good model! where are all the Phi-2 finetunes? Here's code and a guide on how to finetune Phi-2 using QLoRA + includes a part on creating a dataset from a seed of instructions

I’m still using a 7B model in production, mainly because it just works well for these tasks and it’s cheap to run. Why would I pay for gpt-4 exactly when I can apply a Lora over a 7B model for a fraction of the cost

Claude 2 is really good indeed. GPT-4's ability to recognize and describe images is impressive. However, Claude 2's capability to work with PDFs might be an even more significant productivity boost. Maybe it's not one vs the other but about using both (for different things).

Simon Willison @simonw

6 months ago

If you haven't tried Claude yet it's a absolutely worth spending time with - I lean on it a lot for working with longer documents, since it can handle 100,000 tokens (GPT-4 is only 8,000) at a time Plus you can upload PDFs to it - I've used it with 100+ page documents

57 94 1K 704K 666

An interesting and enjoyable read from Léon Bottou and Bernhardt Schölkopf. It suggests different analogies and metaphors for framing what's going on with large language models through the imagery of Jorge Luis Borge, e.g: Fiction Machines & Vindications arxiv.org/abs/2310.01425