[compressor] Codes for post-training quantization & pruning

[csyhhu]

Quantization

• 4-bit quantization: https://github.com/submission2019/cnn-quantization

Pruning

• You may try pruning by absoluted value: Learning both Weights and Connections for Efficient Neural Networks