Great collaborations yield amazing work! Google + Snorkel AI join forces again, this time in LA to share best practices on fine tuning, prompting and AI data development to yield better model results.
I've gone to a few SoCal GenAI Meetups, and I'm happy to say Snorkel AI contributed to another successful one last night at Google's office in LA. We had over 100 attendees (highest yet!). Standing room only for Bradley Fowler's talk on LLM fine tuning. Brad got in a rhythm going through notebook code on advanced prompt designs, labeling functions (LFs), quality models. Even broke out some theory breaking down SFT, PPO, DPO, KTO, ORPO (!!!!) (IYKYK). Absolute ML savagery. And - surprisingly - the audience could hang! Thanks to John Carnahan, Rohani Rajaratnam (PMP), Debbie (Ka) Chamkasem for your great work community building here in Southern California. Daniel Dawson Joni Ricard for being incredible hosts from Google. Plus the 9 (!) Snorkelers who came through. 🙏