There’s nothing to read.

Model A: A_1, …, A_n Model B: B_1, …, B_n

C_i = A_i * p + B_i * (1 - p)

In other words, it’s just a linear combination of the other models’ weights, per position.