Akash Srivastava & Mairin “Mo” Duffy join the Ask Noah Show to talk about Red Hat's approach to AI models.
-- During The Show --
00:45 Intro
- Talking about christmas gifts
- Presentation season
- Christmas time
03:30 Communicating Technical Things
- Who are you talking to
- Willingness to learn
- Knowledge for Knowledge sake
- Steve's process for teaching
- We all compare to things we know
- Building vocabulary
- How to know an idea isn't landing
- What do you aim for time wise
10:28 Red Hat Granite Models
- Small Language Model
- Language and Code Model
- Indemnify the Model
- What is an Open Source Model?
- Models are not software
- Weights
- Why smaller models for enterprise use?
- Tuning models
- Why monolithic vs agentic AI?
- Is it better to train your own model?
- Beam search and inference time scaling
- Why InstructLab?
- InstructLab safety rails
- HuggingFace, InstructLab, formats
- AI in general use
-- The Extra Credit Section --
For links to the articles and material referenced in this week's episode check out this week's page from our podcast dashboard!
This Episode's Podcast Dashboard
Phone Systems for Ask Noah provided by Voxtelesys
Join us in our dedicated chatroom #GeekLab:linuxdelta.com on Matrix
-- Stay In Touch --
Find all the resources for this show on the Ask Noah Dashboard
Need more help than a radio show can offer? Altispeed provides commercial IT services and they’re excited to offer you a great deal for listening to the Ask Noah Show. Call today and ask about the discount for listeners of the Ask Noah Show!
Contact Noah
live [at] asknoahshow.com
-- Twitter --