Pytorch Code积累

2017 Python最新面试题及答案16道题 

15个重要Python面试题 测测你适不适合做Python?

torch.squeeze()

Returns a tensor with all the dimensions of input of size 1 removed.

torch.unsqueeze(input, dim, out=None) → Tensor
Returns a new tensor with a dimension of size one inserted at the specified position.

Python 3:filter()

filter() 函数用于过滤序列,过滤掉不符合条件的元素,返回一个迭代器对象,如果要转换为列表,可以使用 list() 来转换。

该接收两个参数,第一个为函数,第二个为序列,序列的每个元素作为参数传递给函数进行判,然后返回 True 或 False,最后将返回 True 的元素放到新列表中。

 

Gensim

Gensim是一款开源的第三方Python工具包,用于从原始的非结构化的文本中,无监督地学习到文本隐层的主题向量表达。它支持包括TF-IDF,LSA,LDA,和word2vec在内的多种主题模型算法,支持流式训练,并提供了诸如相似度计算,信息检索等一些常用任务的API接口。简单地说,Gensim主要处理文本数据,对文本数据进行建模挖掘。

https://blog.csdn.net/HuangZhang_123/article/details/80326363

traceback

捕获并打印异常,可以输出哪个文件哪个函数哪一行报的错。

@classmethod,@staticmethod,@property

 

 torch.max()

 __call__

在类中实现该方法,一个类实例可以变成一个可调用对象。

代码出处:https://www.cnblogs.com/superxuezhazha/p/5793536.html

更多特殊函数: https://www.cnblogs.com/xiao987334176/p/8884002.html#autoid-0-1-0

 

IoU(Intersection over Union)的计算

 def IOU(xywh1, xywh2):
    x1, y1, w1, h1 = xywh1
    x2, y2, w2, h2 = xywh2

    dx = min(x1+w1, x2+w2) - max(x1, x2)
    dy = min(y1+h1, y2+h2) - max(y1, y2)
    intersection = dx * dy if (dx >=0 and dy >= 0) else 0.
    
    union = w1 * h1 + w2 * h2 - intersection
    return (intersection / union)

其中(x1,y1),(x2,y2)分别为两个矩阵左下角的顶点,w,h为宽和高。

 

xml解析

https://www.cnblogs.com/zqchen/articles/3936805.html

 

layer of model

model.children() returns an iterable of high-level layers present in model.

model.named_children() returns an iterable of two-element tuples, where the first element is the name of the high-level layer and the second element is the high-level layer.

inception loss

           if is_inception and phase == 'train':
                        # From https://discuss.pytorch.org/t/how-to-optimize-inception-model-with-auxiliary-classifiers/7958
                        outputs, aux_outputs = model(inputs)
                        loss1 = criterion(outputs, labels)
                        loss2 = criterion(aux_outputs, labels)
                        loss = loss1 + 0.4*loss2

inception_v3 requires the input size to be (299,299), whereas all of the other models expect (224,224).

 

详解Pytorch中的网络构造(nn.Module)

https://zhuanlan.zhihu.com/p/53927068

 

affine_grad和grid_sample

https://www.jianshu.com/p/723af68beb2e

 

torch.gather

torch.cat

.view() : reshape a tensor.

 By default, user created Tensors have 'requires_grad = False'

.requires_grad_() 和 .detach()

torch.nn.ReplicationPad2d(padding)

Pads the input tensor using replication of the input boundary.

torch.tensor(np_array):

 

.numpy():Converting a Torch Tensor to a NumPy Array

.from_numpy: Converting NumPy Array to Torch Tensor

tensor_b is a different view (interpretation) of the same data present in the underlying storage

torch.stack: 增加新的维度做堆叠

 

 torch.masked_select :在训练阶段,损失函数通常需要进行mask操作,因为一个batch中句子的长度通常是不一样的,一个batch中不足长度的位置需要进行填充(pad)补0,最后生成句子计算loss时需要忽略那些原本是pad的位置的值,即只保留mask中值为1位置的值,忽略值为0位置的值

 

Python标准库(3.x): itertools库

https://www.cnblogs.com/tp1226/p/8453564.html

Python标准库(3.x): 内建函数

https://www.cnblogs.com/tp1226/p/8446503.html

 

torch.einsum:

https://www.jqr.com/article/000481

 

Scikit-Learn中TF-IDF权重计算方法主要用到两个类:CountVectorizerTfidfTransformer

 

select first/last N


select by specific index

index_select


select by mask 会把数据打平












 

select by flatten index



 

 

 

Tensor维度变换

view/reshape: 丢失维度信息

 

squeeze

unsqueeze:插入的index的取值范围[-a.dim()-1, a.dim()+1)

 

transpose / .t() (only for 2D)

 

permute

expand: broadcasting (推荐)

repeat: memory copied

 

Broadcast

    Expand

 without copying data

   match from last dim

  

Merge or Split

cat: concate的维度可以不一样,其它维度必须一样.

 

 stack: create new dim at the dim value

所有的维度必须一样(e.g. a和b).

 

split: by length

chunk: by num

 

*: element-wise

matmul:matrix 乘法, torch.mm(only for 2D)、torch.matmul、@

>2d tensor matmul

.floor() .ceil()

.round(): 四舍五入

.trunc() .frac()

 

clamp: gradient clipping, (min), (min, max)

 

statistics

norm: 范数用来衡量一个向量的大小

mean sum

prod

max, min,

argmin, argmax: 返回的是索引

dim, keepdim

kthvalue(返回第k小), topk

 

compare

torch.eq

torch.gt

 

Tensor advanced operation

where   (condition, a, b)

gather

 

Image Preprocessing:

Image Resize

Data Augumentation

Normalize

ToTensor

 

Numpy中数组索引为None

 

 

 

 

https://www.kaggle.com/gyani95/380000-lyrics-from-
metrolyrics

posted @ 2018-10-30 19:06  一窍不通  阅读(1332)  评论(0编辑  收藏  举报