Top 11 cutting-edge AI/ML things I want to do

  1. Train a LORA adapter for LLM on my mac, adapting massive models for my use without thousands of dollars in GPU dollars. Use it as an english teacher.

  2. Play around with a local visual generative model and see what interesting creations I can come up with without using the frustrating Stable diffusion online that's started taking forever to run.

  3. Use candle to turn standard models into 'serverless' deployments.

  4. Use open-source models as my personal assistant for everyday tasks, a CLI-based guide to help me with various things, all for free.

  5. Use NERF models to generate 3d models and 3d prints based on point-cloud generation from an image/video from a regular phone camera

  6. Actually find good uses for LLM models to improve everyday efficiency for personal gain (or professional gain) that is measurable

  7. Start a consulting company that installs local LLM's for companies wanting to explore AI/ML without investing in large infrastructure.

  8. Write a lot on AI/ML stuff and be a 'domain expert' so I can use it to get other gigs.

  9. Actually use AI/ML for one of my projects, to generate a new idea or something useful that's not super lame. When I tried chatGPT to generate topics for my podcast they were super...how shall I say...lowbrow.

  10. Quantized models. What are they? How to use them? Figure it all out!

  11. Understand the weird data structures, such as bloom embedding used for ML/AI

No comments:

Post a Comment

Tell me what you think. I'll read, promise.