dorefa 的梯度量化问题

666DZY666 / micronet

micronet, a model compression and deploy lib. compression: 1、quantization: quantization-aware-training(QAT), High-Bit(>2b)(DoReFa/Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference)、Low-Bit(≤2b)/Ternary and Binary(TWN/BNN/XNOR-Net); post-training-quantization(PTQ), 8-bit(tensorrt); 2、 pruning: normal、regular and group convolutional channel pruning; 3、 group convolution structure; 4、batch-normalization fuse for quantization. deploy: tensorrt, fp32/fp16/int8(ptq-calibration)、op-adapt(upsample)、dynamic_shape

MIT License

2.2k stars 478 forks source link

dorefa 的梯度量化问题 #62

Open prio158 opened 3 years ago

prio158 commented 3 years ago

你好，首先很感谢这份优秀的工作！我好像在代码中没有看到关于dorefa 的梯度量化的部分，请问是否缺失呢？

sertreet commented 3 years ago

赞同，我也没有看到

666DZY666 commented 3 years ago

梯度目前没有量化