Where Meta stands in the AI race

Good Morning AI Runners

Here's what we've got for you today:

• 2 significant research papers on artificial intelligence to read this weekend.

• OPT-IML: a new language model from Meta AI

• Pic of the day

2 significant research papers on artificial intelligence to read this weekend.

If you ask me what i’ll be spending my reading time on this weekend, it’ll be these two research papers:

• Attention Is All You Need (link): The Transformer model is based on this study, which marks a significant improvement in the application of the attention mechanism. (to dumb it down a bit, "Transformer" is a model that accelerates the rate at which different models may be taught by using "attention").

• Computing Machinery and Intelligence (link): This is one of the classics where Alan Turing addresses the problems of Artificial Intelligence, and proposes an experiment that became known as the Turing test. Turing begins with this question: "Can machines think?"

Even if these research papers seem too difficult to grasp at first, we recommending reading through them if you are interested in learning more about AI.

Where Meta stands in the AI race

Meta AI just announced OPT-IML, a new language model from Meta AI with 175 billion parameters, fine-tuned on 2,000 language tasks.

Let's dumb this down...

• Meta has an AI department (not just VR) and recently announced the release of a new language model which has a hell of a long name: "Open-Pre-trained-Transformer - Instruction Meta-Learning” or just OPT-IML.

• This 175-billion parameter language model was fine-tuned on 2,000 language tasks: trained using natural language instructions in order to improve the model's processing of our own natural language. Natural language processing (NLP) tasks are given/taught to the model as a type of command or instruction (example in the image below).

• Meta is releasing the 30-billion parameter version of the model on Github soon (request form)

Example of NLP Task and Natural Language Instructions (source)

Not gonna lie, this was a heavy post today! but now we leave you with the PPPs:(something to pick up, project to pilot, person to watch)

Run The AI: PPPs

Pick up (learn):

Great video explaining the research paper "Attention Is All You Need" we mentioned above:

Attention Is All You Need

Pilot (play):

DAIR.AI's Prompt Engineering Guide: This guide contains a non-exhaustive set of learning guides and tools about prompt engineering. It includes several materials, guides, examples, papers, and much more. This repo is intented to be used as a research and educational reference for practitioners and developers.

Person:

Our person of the day is Emad Mostaque, founder of Stable Diffusion.

Pic of the day:

That's it from RuntheAI for today.

THANK YOU FOR READING AND SEE YOU TOMORROW, SUBSCRIBE TO STAY UPDATED!