Gradient Descent: Linear Fit & Loss Surface

Initial slope m₀

Initial intercept c₀

Learning rate α

Steps

Noise σ

True m

True c

Step

Loss L

–

(m, c)

–

Click the loss heat map to set the starting point (m₀, c₀). Use True m and True c to change the ground-truth line that generates the data. Adjust Noise σ and hit Resample for fresh data with the same truth.