AI Summit_Sept. 13 2024

Hypo 3: Map it!

Smarter Foundation Model (still GPT)

Baseline Foundation Model

Synthetic data from InstructLab process

Training Data (Pretraining)

RAG

Tuned Model

Database or Search Engine Current Information

•

Custom training Full fine-tuning

•

•

Reinforcement learning Parameter Efficient Fine tuning • Prompt engineering

•

Prompt

Output

14

Practical Takeaways for Resolving Complex Dilemmas

AI Roundtable Page 667