Find out!

123456789101112
Across
  1. 6. m×n matrix M. what is nxn matrix called?
  2. 7. ignores all but the most “active”
  3. 9. number of batches needed to complete one epoch
  4. 10. gradient of a function at saddle point
  5. 11. measures how good you are
  6. 12. ranges from ZERO to INFINITY
Down
  1. 1. layers which learn spatial features in CNN
  2. 2. low bias makes models to ..........
  3. 3. assigned prior to training
  4. 4. opposite of vanish
  5. 5. Word embedding technique
  6. 7. solves problems of border effects
  7. 8. I can improve or decrease the updates