Don't have Telegram yet? Try it now!
https://hgpu.org/?p=16582
Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations