hi, 参数：last|activation_mp|activation_nor|last_relu —— 模型结构调整标志位,这些参数都是什么意思呢？🙏

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

2.21k stars 478 forks source link

hi, 参数：last|activation_mp|activation_nor|last_relu —— 模型结构调整标志位,这些参数都是什么意思呢？🙏 #11

Open HaoMyWorld opened 4 years ago

666DZY666 commented 4 years ago

都是对量化模型结构的调整标志位； last | last_relu：模型最后一层卷积的A（输入特征）是否二值 activation_mp | activation_nor：保证C-B-A-C-B-A-C-B-A-MP —> C-B-A......的网络结构顺序

HaoMyWorld commented 4 years ago

hi, util_w_t_b中的卷积阈值2/3值的公式有什么依据或者论文吗？

666DZY666 commented 4 years ago

1、现在这些标识位大都简化去掉了 2、相关论文已补充