qq_45759229的博客

center loss pytorch实现总结

pytorch ignite使用

pytorch_lightning使用教程

python将列表元素解释为函数参数

tensorfow与pytorch进行转换

cuda out of memory (epoch=120)时发生

cosine相似度计算总结

我测试发现，keras在比较简单的网络上远比pytorch跑的要快的多，目前大致的猜测原因在读取数据部分，但是我还不知道怎么pytorch这个数据的读取速度，为什么都没有大佬发现这个问题呢？mnist数据集读取import osos.environ["CUDA_VISIBLE_DEVICES"] = "0"import torchfrom torchvision import datasets, transformsimport matplotlib.pyplot as pltimport t

import torch torch.manual_seed(0)print(torch.randn(1, 2))print(torch.randn(1, 2))print("======================") # torch.manual_seed()控制torch中随机数的生成，并不控制numpy和random中torch.manual_seed(0)print(torch.randn(1, 2))print(torch.randn(1, 2))然后就有一个好玩的事

对于pytorch的结果复现，对于普通的设置使用下面的函数即可参考https://stackoverflow.com/questions/60386708/random-seed-for-order-of-elements-in-pythons-set-to-list-conversionpytorch复现结果def seed_torch(seed=42): random.seed(seed) os.environ['PYTHONHASHSEED'] = str(seed)

pytorchimport numpy as npimport torch.nn as nnimport randomimport os import torchimport seaborn as snsimport matplotlib.pyplot as pltdef seed_torch(seed=42): random.seed(seed) os.environ['PYTHONHASHSEED'] = '0' np.random.seed(seed)

tensorflowimport tensorflow.keras.backend as Kfrom tensorflow.keras.callbacks import TensorBoard, ModelCheckpoint, EarlyStopping, ReduceLROnPlateau,Historyfrom tensorflow.keras.layers import Dense, Input,Layer,InputSpecfrom tensorflow.keras.models impo

earlystopping实现https://github.com/Bjarten/early-stopping-pytorch自己实现LinearRegression# 我通过一个线性回归(或者一个自编码器)来联系使用earlystopimport torchimport numpy as npimport torch.nn as nn#X1=np.linspace(-10,10,100)#X2=np.linspace(-20,20,100)#X3=np.linspace(0,40,1

下面进行代码验证单个样本import torch.nn.functional as F# this is the same example in wikiP = F.softmax(torch.randn(1,4),dim=1)Q = F.softmax(torch.randn(1,4),dim=1)print((P * (P / Q).log()).sum()) # # tensor(0.0863), 10.2 µs ± 508print(F.kl_div(Q.log(), P, r

import torchimport torchvisionfrom torch import nnfrom torch.autograd import Variablefrom torch.utils.data import DataLoaderfrom torchvision import transformsfrom torchvision.utils import save_imagefrom torchvision.datasets import MNISTimport osba

https://zhuanlan.zhihu.com/p/361431647https://zhuanlan.zhihu.com/p/345450458import torchfrom torch import nnembedding = nn.Embedding(5, 4) # 假定字典中只有5个词，词向量维度为4word = [[1, 2, 3], [2, 3, 4]] # 每个数字代表一个词，例如 {'!':0,'how':1, 'are':2, 'you':3, 'ok

参考https://zhuanlan.zhihu.com/p/35434175https://www.kaggle.com/scottclowe/testing-gpu-enabled-notebooks-mnist-pytorch#%%time import torchimport torch.nn as nnimport torch.nn.functional as Fimport numpy as npimport matplotlib.pyplot as pltfrom torch

import torch.nn as nnimport torchloss = nn.L1Loss()a=torch.FloatTensor([1,2,3])b=torch.FloatTensor([1.1,2.2,2.9]) output = loss(a, b)print(output)c=torch.FloatTensor([[1,3],[2,3],[1,1]])d=torch.FloatTensor([[2,3.2],[2.1,3.2],[1.1,1.1]])out=loss(c

kaggle uspsimport numpy as np # linear algebraimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)import os#print(os.listdir("../input"))import h5py from functools import reducedef hdf5(path, data_key = "data", target_key = "targe