【经济学】量化模型Qlib图片速览 “Qlib: An AI-oriented Quantitative Investment Platform” 论文阅读

PCM （脉冲编码调制）

ONNXhttps://onnx.ai/GitHub：Open standard for machine learning interoperability官方文档概述支持的算子导出模型：pytorch官方的例子(可以看到将 device设置为"cpu"，方便兼容读者进行测试)：import torchimport torchvisiondummy_input = torch.randn(10,

http://playground.tensorflow.org/源码

import Read_RectTXTimport numpy as npimport torchvision.transforms as transformsimport cv2 as cvfilename = 'DATA/groundtruth_rect.txt'rect = Read_RectTXT.Read_RectTXT(filename)img_iter = 1img = cv.imr

让每一层的输出和梯度均值为零方差固定的随机biaanling让每一层的输出和梯度均值为零方差固定的随机biaanling让每一层的输出和梯度均值为零方差固定的随机biaanlingiid :独立通分布即若输入的方差和输出的方差一样的话：nt−1∗γt=1即若输入的方差和输出的方差一样的话：n_{t-1} * γ_t=1即若输入的方差和输出的方差一样的话：nt−1​∗γt​=1反向类似：其中nt−1

x =torch.tensor([[[1,2,3],[1,1,0]],[[1,1,0],[1,1,0]]])# torch.Size([2, 2, 3])y= torch.tensor([[1],[1],[0]]) #torch.Size([3, 1])res= torch.matmul(x, y)# torch.Size([2, 2, 1])tensor([[[3], [2]],[[2],[2]

CPP项目中使用CUDA(opencv+cuda or pybind+cuda)

opencv图像增强操作

# 一切皆对象let a:object;var a = {};# 声明+初始化var a= {key1: "key1",key2: 123,key3: function() {return "key3";},}# 先声明再初始化let b= {key1:string,}b = { key1:"key1"}# ?: 可选属性(初始化时可以没有)let c= {key1:string ,