The length of pytorch dataloader the difference between epoch and iteration

pytorch dataloader The length of

pytorch in dataloader The size of will depend on batch_size Automatic resizing of . If the training dataset has 1000 Samples , also batch_size The size is 10, be dataloader The length of 100.

It should be noted that , If dataset The size of cannot be batch_size to be divisible by , be dataloader Last of batch Maybe more than the actual batch_size smaller . for example , about 1001 Samples ,batch_size Its size is 10,train_loader The length of len(train_loader)=101, the last one batch Will contain only one sample . Can be set by dataloader Of drop_last The attribute is True To avoid this .

Pytorch Of Dataset Provides access to dataset Interface to a single sample in , By using index Subscript to get single sample data .Dataloader To provide with batch Samples for model training
 

epoch And iteration The difference between

iteration: 1 individual iteration Equal to using batchsize One sample training

epoch: 1 individual epoch It is equal to training once with all the samples in the training set