Performance of Post-Training Two-Bits Uniform and Layer-Wise Uniform Quantization for MNIST Dataset from the Perspective of Support Region Choice

<div>Normalized histogram of all weights (FP32); transfer characteristic of the symmetric two-bits UQ; merged normalized histogram of weights FP32 and uniformly quantized (a) Case 4 (<i>x</i><sub>max</sub><sup>[J]</sup> = 2.1748) and (b) Case 1 (<i>x</i><sub>max</sub> = 4.8371024).</div>

Mathematical Problems in Engineering

fig5

Figure 5

Figure 5: Performance of Post-Training Two-Bits Uniform and Layer-Wise Uniform Quantization for MNIST Dataset from the Perspective of Support Region Choice 

Figure 5 | Performance of Post-Training Two-Bits Uniform and Layer-Wise Uniform Quantization for MNIST Dataset from the Perspective of Support Region Choice 