05 Checkpoint: Optimisation and Numerical Methods

Content and key concepts

Progress is available with JavaScript. View is not equal to development.
Eight reviews of the mechanism
Threshold: 7 out of 8. without a timer. Balls are given for answers, not for visits. This is a local self-check; the decisions will be shown after dispatch.
1. The coordinates of the loss Landscape point here mean...
Model Options
Token numbers
Color image channels
2. A big lightning rate can...
Just speed up.
Increase the loss and cause input
Make Gradient exactly right.
3. Medium gradiant minibatch...
Always equals total
May be different from total
It's a random number without data.
4. The second moment gradient is equal to dispersion?
Yeah, always.
No, we need another square of average for variance.
On GPU only
5. The first EMA momentum m prig=4,β=9,m0=0?
Your numerical response
6. Full gradient for theta 0,targets[1,3.5]?
Your numerical response
7. First coordinate norm-clipped [3.4] ac = 2?
Your numerical response
8. The updated "decay:,2,1,5,"?
Your numerical response
The result will appear after all the answers have been sent.
Research and explain
Compare the methods with one budget, then change the concentration. Do not conclude on the quality of LLM from the same graph. Show on the state of step one Adam, where the initial shift disappears; explain separately why clipping and decay are not one action.
First, shape the hypothesis, then change one parameter. Compare the numbers to the formula; state the extreme case and the limit of applicability. The free conclusion is not evaluated by a dummy automatic score.