FOR IMMEDIATE RELEASE: 17/Jul/2022
Welcome back to The Memo.
These emails now come from a new platform with better access to archives, simpler handling of credit card expiration and renewal, cleaner layout, faster publishing, and maybe even video backups.
The BIG Stuff
New article: Use cases for large language models including GPT-3 (15/Jul/2022)
Here are some of the major achievements and use cases of large language models (LLMs) like GPT-3. This list is not exhaustive, there are many, many use cases for large language models in 2022 and beyond. As the creator of LLM Aleph Alpha Luminous says:
“[LLMs can] Solve every problem that can be expressed with language.”
Read the article: https://lifearchitect.ai/use-cases/
Watch the video:
Meta AI releases NLLB (Jul/2022)
Meta AI has released NLLB (no language left behind), a 54.5B MoE model and a 3.3B dense model, covering 200+ languages.
“Driven by the goal of eradicating language barriers on a global scale, machine translation has solidified itself as a key focus of artificial intelligence research today. However, such efforts have coalesced around a small subset of languages [like English, Spanish, and French], leaving behind the vast majority of mostly low-resource languages. What does it take to break the 200 language barrier while ensuring safe, high quality results, all while keeping ethical considerations in mind? In No Language Left Behind, we took on this challenge by first contextualizing the need for low-resource language translation support through exploratory interviews with native speakers. Then, we created datasets and models aimed at narrowing the performance gap between low and high-resource languages.”
Announce: https://ai.facebook.com/research/no-language-left-behind/
And the beautiful demo: https://nllb.metademolab.com/
The original paper is 190 pages long(!), and includes a table of all languages used in training. Although, just as a personal side note, Meta AI seems to be missing the 363 Indigenous Australian languages (map below shows just a subset of these).
The Interesting Stuff
Exclusive first look: GPT-3 + DALL-E 2 + VLM (17/Jul/2022)
I think this is a world first, but I don’t mind being corrected! In this special edition, Leta AI writes a children’s book, and then its prompts are used to generate AI images via DALL-E 2. Lastly, we use Quickchat.ai Emerson and its visual language model to look at the images and comment on them. I had a lot of fun with this, and as a paid reader of The Memo, you get to see it first AND you get a complimentary copy of the eBook!
Watch the video:
Download the book: