Size tilt in a factor model

Could someone please provide clarity on whether a negative size coefficient (in a factor model) represents a large cap or small cap tilt? Please could you explain your answer as well.

R29 Example 7 CFAI text shows a negative size coefficient and refers to it as a small cap tilt

R29 Example a negative (more negative than the benchmark) and refers to it as a large cap tilt

OK

If size coefficient = Small Caps - Large Caps, negative coefficient represents a large cap tilt.

If size coefficient = Large - Small Caps, negative coefficient represents a small cap tilt.

Simple’s that.