Previous Review:
- Before the model create a mini-P&L + after then review the AUC and blend the discussion
Much of this is not new
- Unsupervised learning still requires features, so our bias exists.
- Could just be a neural network that’s running a linear regression
Equations:
- Tell me what the outcome is…
- What’s data and what’s not data?
- What on this is data and what’s an outcome
Neural Networks Mechanics
- Coefficients, parameters and weights
- Interaction effect
- Deep learning —> original excel spread sheet w/ 3 columns and now you have 10 columns. New data into my model
The Unlock
- GPU (graphical processing units); massive data and large processing power.
Difference between math and language: Where you want certainty and creativity. The breakthrough for LLMs require context. When it can unlock context
The models are tested and trained on english which will further increase those languages most commonly used
AI should make you ask not, “what can I automate?” Instead, “do you even want to be in this business anymore?”