Tweeted By @TensorFlow
Post-training integer quantization is now available in the TensorFlow Model Optimization Toolkit!
— TensorFlow (@TensorFlow) June 11, 2019
Quantized models can be 2-4x faster on CPU, 4x smaller with low accuracy loss, and are compatible with fixed-point accelerators.
Learn more here ↓ https://t.co/R4w3ldaRVc