Episode Details
Back to Episodes
๐ ThursdAI Nov 16 - Live AI art, MS copilots everywhere, EMUs from Meta, sketch-to-code from TLDraw, Capybara 34B and other AI news!
Description
Hey yall, welcome to this special edition of ThursdAI. This is the first one that I'm sending in my new capacity as the AI Evangelist Weights & Biases (on the growth team)
I made the announcement last week, but this week is my first official week at W&B, and oh boy... how humbled and excited I was to receive all the inspiring and supporting feedback from the community, friends, colleagues and family ๐โโ๏ธ
I promise to continue my mission of delivering AI news, positivity and excitement, and to be that one place where we stay up to date so you don't have to.
This week we also had one of our biggest live recordings yet, with 900 folks tuned in so far ๐ฎ and it was my pleasure to again to chat with folks who "made the news" so we had a brief interview with Steve Ruiz and Lou from TLDraw, about their incredible GPT-4 Vision enabled "make real" functionality and finally got to catch up with my good friend Idan Gazit who's heading the Github@Next team (the birthplace of Github Copilot) about how they see the future. So definitely definitely check out the full conversation!
TL;DR of all topics covered:
* Open Source LLMs
* Nous Capybara 34B on top of Yi-34B (with 200K context length!) (Eval, HF)
* Microsoft - Phi 2 will be open sourced (barely) (Announcement, Model)
* HF adds finetune chain genealogy (Announcement)
* Big CO LLMs + APIs
* Microsoft - Everything is CoPilot (Summary, copilot.microsoft.com)
* CoPilot for work and 365 (Blogpost)
* CoPilot studio - low code "tools" builder for CoPilot + GPTs access (Thread)
* OpenAI Assistants API cookbook (Link)
* Vision
* ๐ฅ TLdraw make real button - turn sketches into code in seconds with vision (Video, makereal.tldraw.com)
* Humane Pin - Orders are out, shipping early 2024, multimodal AI agent on your lapel (
* )
* Voice & Audio
* ๐ฅ DeepMind (Youtube) - Lyria high quality music generations you can HUM into (Announcement)
* EmotiVoice - 2000 different voices with emotional synthesis (Github)
* Whisper V3 is top of the charts again (Announcement, Leaderboard,
Listen Now
Love PodBriefly?
If you like Podbriefly.com, please consider donating to support the ongoing development.
Support Us