Model Training & Adaptation
Gradient descent
An optimization method that updates parameters in the direction that reduces loss.
Example
The optimizer takes repeated downhill steps on the loss surface.
Why it matters
It gives training a practical way to improve internal settings through repeated adjustments.
You'll hear it
“The search is improving, but the steps may be too large.”
Can you control it?
Set by the team
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Does it always find the best possible solution?
- No. The path and final result depend on the problem and starting choices.
- Can step size affect whether it settles down?
- Yes. Steps that are too large can overshoot, while very small steps may progress slowly.
- Does downhill refer to a physical movement?
- No. It is a visual description of moving toward a lower error score.