LLMOps Concepts
Max Knobbout, PhD
Applied Scientist, Uber
Resulting in what we call LLM applications throughout this course
Some differences:
LLMOps | MLOps | |
---|---|---|
Model size | Large | Typically smaller |
Data | Text | Any data |
Pre-trained models | Typically yes | Typically no |
Model improvement | Prompt engineering & fine-tuning | Feature engineering & model selection |
Generalization | General-purpose | Fixed scope |
Unpredictability | High | Low |
Output | Primarily text | Task-specific |
LLMOps Concepts