【Pytorch】量化

#-*- coding:utf-8 -*-
import torch
 
#量化仅可用cpu
model = ResNet().cpu()
model = torch.load_state_dict(torch.load(weights))
 
#Specify quantization configuration
#在这一步声明了对称量化或非对称量化，及量化bit数
#如下代码中采用了默认的非对称量化，及8bit
model.qconfig = torch.quantization.default_qconfig
model = torch.quantization.prepare(model)
 
#Convert to quantized model
model = torch.quantization.convert(model)
 
#Save model, 保存后模型的size显著减小，但性能损失相对较大
#故，建议考虑量化感知训练
torch.save(model.state_dict(), "path.pt")

https://blog.youkuaiyun.com/perfects110/article/details/108804622?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522165085165816782388088547%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fall.%2522%257D&request_id=165085165816782388088547&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2_allfirst_rank_ecpm_v1~rank_v31_ecpm-1-108804622.142^v9control,157^v4control&utm_term=%E5%9F%BA%E4%BA%8Epytorch%E7%9A%84%E6%A8%A1%E5%9E%8B%E9%87%8F%E5%8C%96%E5%AE%9E%E7%8E%B0&spm=1018.2226.3001.4187
https://cloud.tencent.com/developer/article/1673489