Kernel and Bias Quantizers not quantizing weights and biases

I trained a qkeras model with kernel and bias quantizers for every QDense layer as `quantized_bits(8,0)`. After training, I print out the weights and biases of the QDense layers.

I expect them to have 7 or 8 bits in their _binary representation_ according to the documentation . However, I find all of them to be exceeding that _by a lot_ .

A sample of the weights that I find upon printing them out:

-1.18689373e-01, 7.44902715e-03, -1.58425614e-01, 7.54895657e-02,
-4.10564430e-03, 2.46057995e-02

Clearly, they do not have 7 or 8 bits in their _binary representation_ .

So, my question is: Are the kernel and bias quantizers not functioning properly? Or am I missing something?