W20D2: Keep track of everything you try!
Write down what choices you made and explain why. What did you try? What worked? What didn't?
| Decision Point | Alternatives Considered | Selected Approach | Justification (cite theory/evidence) |
|---|---|---|---|
| Algorithm Type | |||
| Network Size | |||
| How It Explores | |||
| Learning Speed |
Buffer size, batch size, update frequency
Update frequency τ or period
Before you start training, write down what you think will happen. This helps you learn - even if you're wrong!
Keep track of what you asked the AI and what it gave you. What worked? What did you have to fix?
| # | What I Asked | How Good Was It? | Problems I Found | What I Changed |
|---|---|---|---|---|
| 1 | ||||
| 2 |
Write down every experiment you ran. Include the ones that didn't work too - you learn just as much from failures!
| Exp ID | Configuration | Episodes | Mean ± Std | Solved? | Wall Time | Notes |
|---|---|---|---|---|---|---|
Look at your results and figure out what they mean. Which settings worked best? How do you know?
| What I Predicted | Result | Supporting Evidence |
|---|---|---|
| H₁ (Primary) | ||
| H₂ (Stabilization) | ||
| H₃ (Hyperparameters) |
Try removing or changing parts of your code to see what happens. This helps you understand which parts actually matter!
| What I Removed/Changed | Score With It | Score Without It | Δ Performance | Interpretation |
|---|---|---|---|---|
Think about what you discovered and what you'd do differently next time.
Write down everything that helped you: websites, videos, tutorials, and AI chats. It's important to give credit!
Paste the links so you can find them again later!