Documentaries (though they often include photos or video portions that can be considered primary sources). Show
When is a Primary Source a Secondary Source?Whether something is a primary or secondary source often depends upon the topic and its use. A biology textbook would be considered a secondary source if in the field of biology, since it describes and interprets the science but makes no original contribution to it. On the other hand, if the topic is science education and the history of textbooks, textbooks could be used a primary sources to look at how they have changed over time. At the heart of PyTorch data loading utility is the class. It represents a Python iterable over a dataset, with support for
These options are configured by the constructor arguments of a , which has signature: DataLoader(dataset, batch_size=1, shuffle=False, sampler=None,
The sections below describe in details the effects and usages of these options.Dataset TypesThe most important argument of constructor is , which indicates a dataset object to load data from. PyTorch supports two different types of datasets:
Map-style datasetsA map-style dataset is one that implements the dataset_iter = iter(dataset) for indices in batch_sampler:
1 anddataset_iter = iter(dataset) for indices in batch_sampler:
2 protocols, and represents a map from (possibly non-integral) indices/keys to data samples.For example, such a dataset, when accessed with dataset_iter = iter(dataset) for indices in batch_sampler:
3, could read the dataset_iter = iter(dataset) for indices in batch_sampler:
4-th image and its corresponding label from a folder on the disk.See for more details. Iterable-style datasetsAn iterable-style dataset is an instance of a subclass of that implements the dataset_iter = iter(dataset) for indices in batch_sampler:
7 protocol, and represents an iterable over data samples. This type of datasets is particularly suitable for cases where random reads are expensive or even improbable, and where the batch size depends on the fetched data.For example, such a dataset, when called dataset_iter = iter(dataset) for indices in batch_sampler:
8, could return a stream of data reading from a database, a remote server, or even logs generated in real time.See for more details. Data Loading Order andFor , data loading order is entirely controlled by the user-defined iterable. This allows easier implementations of chunk-reading and dynamic batch size (e.g., by yielding a batched sample at each time). The rest of this section concerns the case with . classes are used to specify the sequence of indices/keys used in data loading. They represent iterable objects over the indices to datasets. E.g., in the common case with stochastic gradient decent (SGD), a could randomly permute a list of indices and yield each one at a time, or yield a small number of them for mini-batch SGD. A sequential or shuffled sampler will be automatically constructed based on the for index in sampler:
3 argument to a . Alternatively, users may use the argument to specify a custom object that at each time yields the next index/key to fetch.A custom that yields a list of batch indices at a time can be passed as the for index in sampler:
8 argument. Automatic batching can also be enabled via for index in sampler:
9 andfor data in iter(dataset):
0 arguments. See for more details on this.Note Neither nor for index in sampler:
8 is compatible with iterable-style datasets, since such datasets have no notion of a key or an index.Loading Batched and Non-Batched Datasupports automatically collating individual fetched data samples into batches via arguments for index in sampler:
9, for data in iter(dataset):
0, for index in sampler:
8, andfor data in iter(dataset):
7 (which has a default function).Automatic batching (default)This is the most common case, and corresponds to fetching a minibatch of data and collating them into batched samples, i.e., containing Tensors with one dimension being the batch dimension (usually the first). When for index in sampler:
9 (default for data in iter(dataset):
class SimpleCustomBatch: def collate_wrapper(batch): inps = torch.arange(10 5, dtype=torch.float32).view(10, 5)
tgts = torch.arange(10 5, dtype=torch.float32).view(10, 5)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset, batch_size=2, collate_fn=collate_wrapper, for batch_ndx, sample in enumerate(loader):
0, the data loader yields batched samples instead of individual samples. for index in sampler:
9 andfor data in iter(dataset):
0 arguments are used to specify how the data loader obtains batches of dataset keys. For map-style datasets, users can alternatively specify for index in sampler:
8, which yields a list of keys at a time.Note The for index in sampler:
9 and for data in iter(dataset):
0 arguments essentially are used to construct a for index in sampler:
8 from . For map-style datasets, the is either provided by user or constructed based on the for index in sampler:
3 argument. For iterable-style datasets, the is a dummy infinite one. See on more details on samplers.After fetching a list of samples using the indices from sampler, the function passed as the for data in iter(dataset):
7 argument is used to collate lists of samples into batches.In this case, loading from a map-style dataset is roughly equivalent with: for indices in batch_sampler:
and loading from an iterable-style dataset is roughly equivalent with:dataset_iter = iter(dataset) for indices in batch_sampler:
A custom for data in iter(dataset):
7 can be used to customize collation, e.g., padding sequential data to max length of a batch. See on more about for data in iter(dataset):
7.Disable automatic batchingIn certain cases, users may want to handle batching manually in dataset code, or simply load individual samples. For example, it could be cheaper to directly load batched data (e.g., bulk reads from a database or reading continuous chunks of memory), or the batch size is data dependent, or the program is designed to work on individual samples. Under these scenarios, it’s likely better to not use automatic batching (where for data in iter(dataset):
7 is used to collate the samples), but let the data loader directly return each member of the object.When both for index in sampler:
9 and for index in sampler:
8 are class SimpleCustomBatch: def collate_wrapper(batch): inps = torch.arange(10 5, dtype=torch.float32).view(10, 5)
tgts = torch.arange(10 5, dtype=torch.float32).view(10, 5)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset, batch_size=2, collate_fn=collate_wrapper, for batch_ndx, sample in enumerate(loader):
0 (default value for for index in sampler:
8 is already class SimpleCustomBatch: def collate_wrapper(batch): inps = torch.arange(10 5, dtype=torch.float32).view(10, 5)
tgts = torch.arange(10 5, dtype=torch.float32).view(10, 5)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset, batch_size=2, collate_fn=collate_wrapper, for batch_ndx, sample in enumerate(loader):
0), automatic batching is disabled. Each sample obtained from the is processed with the function passed as the for data in iter(dataset):
7 argument.When automatic batching is disabled, the default for data in iter(dataset):
7 simply converts NumPy arrays into PyTorch Tensors, and keeps everything else untouched.In this case, loading from a map-style dataset is roughly equivalent with: for index in sampler:
and loading from an iterable-style dataset is roughly equivalent with:for data in iter(dataset):
See on more about for data in iter(dataset):
7.Working withfor data in iter(dataset):
7The use of for data in iter(dataset):
7 is slightly different when automatic batching is enabled or disabled.When automatic batching is disabled, for data in iter(dataset):
7 is called with each individual data sample, and the output is yielded from the data loader iterator. In this case, the default for data in iter(dataset):
7 simply converts NumPy arrays in PyTorch tensors.When automatic batching is enabled, for data in iter(dataset):
7 is called with a list of data samples at each time. It is expected to collate the input samples into a batch for yielding from the data loader iterator. The rest of this section describes the behavior of the default for data in iter(dataset):
7 ().For instance, if each data sample consists of a 3-channel image and an integral class label, i.e., each element of the dataset returns a tuple > images = ImageDataset()texts = TextDataset()tuple_stack = StackDataset(images, texts) tuple_stack[0] == (images[0], texts[0]) dict_stack = StackDataset(image=images, text=texts) dict_stack[0] == {'image': images[0], 'text': texts[0]} 2, the default for data in iter(dataset):
7 collates a list of such tuples into a single tuple of a batched image tensor and a batched class label Tensor. In particular, the default for data in iter(dataset):
7 has the following properties:
Users may use customized for data in iter(dataset):
7 to achieve custom batching, e.g., collating along a dimension other than the first, padding sequences of various lengths, or adding support for custom data types.If you run into a situation where the outputs of have dimensions or type that is different from your expectation, you may want to check your for data in iter(dataset):
7.Single- and Multi-process Data LoadingA uses single-process data loading by default. Within a Python process, the Global Interpreter Lock (GIL) prevents true fully parallelizing Python code across threads. To avoid blocking computation code with data loading, PyTorch provides an easy switch to perform multi-process data loading by simply setting the argument > def collate_tensor_fn(batch, *, collate_fn_map): 2 to a positive integer. Single-process data loading (default)In this mode, data fetching is done in the same process a is initialized. Therefore, data loading may block computing. However, this mode may be preferred when resource(s) used for sharing data among processes (e.g., shared memory, file descriptors) is limited, or when the entire dataset is small and can be loaded entirely in memory. Additionally, single-process loading often shows more readable error traces and thus is useful for debugging. Multi-process data loadingSetting the argument > def collate_tensor_fn(batch, *, collate_fn_map): 2 as a positive integer will turn on multi-process data loading with the specified number of loader worker processes. Warning After several iterations, the loader worker processes will consume the same amount of CPU memory as the parent process for all Python objects in the parent process which are accessed from the worker processes. This can be problematic if the Dataset contains a lot of data (e.g., you are loading a very large list of filenames at Dataset construction time) and/or you are using a lot of workers (overall memory usage is > def collate_tensor_fn(batch, *, collate_fn_map): 5). The simplest workaround is to replace Python objects with non-refcounted representations such as Pandas, Numpy or PyArrow objects. Check out for more details on why this occurs and example code for how to workaround these problems. In this mode, each time an iterator of a is created (e.g., when you call > def collate_tensor_fn(batch, *, collate_fn_map): 7), > def collate_tensor_fn(batch, *, collate_fn_map): 2 worker processes are created. At this point, the , for data in iter(dataset):
7, and for indices in batch_sampler:
01 are passed to each worker, where they are used to initialize, and fetch data. This means that dataset access together with its internal IO, transforms (including for data in iter(dataset):
returns various useful information in a worker process (including the worker id, dataset replica, initial seed, etc.), and returns class SimpleCustomBatch: def collate_wrapper(batch): inps = torch.arange(10 5, dtype=torch.float32).view(10, 5)
tgts = torch.arange(10 5, dtype=torch.float32).view(10, 5)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset, batch_size=2, collate_fn=collate_wrapper, for batch_ndx, sample in enumerate(loader):
0 in main process. Users may use this function in dataset code and/or for indices in batch_sampler:
01 to individually configure each dataset replica, and to determine whether the code is running in a worker process. For example, this can be particularly helpful in sharding the dataset.For map-style datasets, the main process generates the indices using and sends them to the workers. So any shuffle randomization is done in the main process which guides loading by assigning indices to load. For iterable-style datasets, since each worker process gets a replica of the object, naive multi-process loading will often result in duplicated data. Using and/or for indices in batch_sampler:
01, users may configure each replica independently. (See documentations for how to achieve this. ) For similar reasons, in multi-process loading, the for data in iter(dataset):
0 argument drops the last non-full batch of each worker’s iterable-style dataset replica.Workers are shut down once the end of the iteration is reached, or when the iterator becomes garbage collected. Warning It is generally not recommended to return CUDA tensors in multi-process loading because of many subtleties in using CUDA and sharing CUDA tensors in multiprocessing (see ). Instead, we recommend using (i.e., setting for indices in batch_sampler:
12), which enables fast data transfer to CUDA-enabled GPUs.Platform-specific behaviorsSince workers rely on Python , worker launch behavior is different on Windows compared to Unix.
This separate serialization means that you should take two steps to ensure you are compatible with Windows while using multi-process data loading:
Randomness in multi-process data loadingBy default, each worker will have its PyTorch seed set to for indices in batch_sampler:
31, where for indices in batch_sampler:
32 is a long generated by main process using its RNG (thereby, consuming a RNG state mandatorily) or a specified for indices in batch_sampler:
33. However, seeds for other libraries may be duplicated upon initializing workers, causing each worker to return identical random numbers. (See in FAQ.).In for indices in batch_sampler:
01, you may access the PyTorch seed set for each worker with either or , and use it to seed other libraries before data loading.Memory PinningHost to GPU copies are much faster when they originate from pinned (page-locked) memory. See for more details on when and how to use pinned memory generally. For data loading, passing for indices in batch_sampler:
12 to a will automatically put the fetched data Tensors in pinned memory, and thus enables faster data transfer to CUDA-enabled GPUs.The default memory pinning logic only recognizes Tensors and maps and iterables containing Tensors. By default, if the pinning logic sees a batch that is a custom type (which will occur if you have a for data in iter(dataset):
7 that returns a custom batch type), or if each element of your batch is a custom type, the pinning logic will not recognize them, and it will return that batch (or those elements) without pinning the memory. To enable memory pinning for custom batch or data type(s), define a for indices in batch_sampler:
40 method on your custom type(s).See the example below. Example: class SimpleCustomBatch: def collate_wrapper(batch): inps = torch.arange(10 5, dtype=torch.float32).view(10, 5)
tgts = torch.arange(10 5, dtype=torch.float32).view(10, 5)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset, batch_size=2, collate_fn=collate_wrapper, for batch_ndx, sample in enumerate(loader):
class torch.utils.data.DataLoader(dataset, batch_size\=1, shuffle\=None, sampler\=None, batch_sampler\=None, num_workers\=0, collate_fn\=None, pin_memory\=False, drop_last\=False, timeout\=0, worker_init_fn\=None, multiprocessing_context\=None, generator\=None, *, prefetch_factor\=None, persistent_workers\=False, pin_memory_device\='') Data loader combines a dataset and a sampler, and provides an iterable over the given dataset. The supports both map-style and iterable-style datasets with single- or multi-process loading, customizing loading order and optional automatic batching (collation) and memory pinning. See documentation page for more details. Parameters
Warning If the for indices in batch_sampler:
79 start method is used, for indices in batch_sampler:
01 cannot be an unpicklable object, e.g., a lambda function. See on more details related to multiprocessing in PyTorch.Warning for indices in batch_sampler:
81 heuristic is based on the length of the sampler used. When is an , it instead returns an estimate based on for indices in batch_sampler:
84, with proper rounding depending on for data in iter(dataset):
0, regardless of multi-process loading configurations. This represents the best guess PyTorch can make because PyTorch trusts user code in correctly handling multi-process loading to avoid duplicate data.However, if sharding results in multiple workers having incomplete last batches, this estimate can still be inaccurate, because (1) an otherwise complete batch can be broken into multiple ones and (2) more than one batch worth of samples can be dropped when for data in iter(dataset):
0 is set. Unfortunately, PyTorch can not detect such cases in general.See for more details on these two types of datasets and how interacts with . class torch.utils.data.Dataset(*args, **kwds) An abstract class representing a . All datasets that represent a map from keys to data samples should subclass it. All subclasses should overwrite dataset_iter = iter(dataset) for indices in batch_sampler:
1, supporting fetching a data sample for a given key. Subclasses could also optionally overwritedataset_iter = iter(dataset) for indices in batch_sampler:
2, which is expected to return the size of the dataset by many implementations and the default options of . Subclasses could also optionally implement for indices in batch_sampler:
94, for speedup batched samples loading. This method accepts list of indices of samples of batch and returns list of samples.Note by default constructs an index sampler that yields integral indices. To make it work with a map-style dataset with non-integral indices/keys, a custom sampler must be provided. class torch.utils.data.IterableDataset(*args, **kwds) An iterable Dataset. All datasets that represent an iterable of data samples should subclass it. Such form of datasets is particularly useful when data come from a stream. All subclasses should overwrite dataset_iter = iter(dataset) for indices in batch_sampler:
7, which would return an iterator of samples in this dataset.When a subclass is used with , each item in the dataset will be yielded from the iterator. When for indices in batch_sampler:
72, each worker process will have a different copy of the dataset object, so it is often desired to configure each copy independently to avoid having duplicate data returned from the workers. , when called in a worker process, returns information about the worker. It can be used in either the dataset’s dataset_iter = iter(dataset) for indices in batch_sampler:
7 method or the ‘sfor indices in batch_sampler:
01 option to modify each copy’s behavior.Example 1: splitting workload across all workers in dataset_iter = iter(dataset) for indices in batch_sampler:
7:> class MyIterableDataset(torch.utils.data.IterableDataset): ... def init(self, start, end): ... super(MyIterableDataset).init() ... assert end > start, "this example code only works with end >= start" ... self.start = start ... self.end = end ... ... def iter(self): ... worker_info = torch.utils.data.get_worker_info() ... if worker_info is None: # single-process data loading, return the full iterator ... iter_start = self.start ... iter_end = self.end ... else: # in a worker process ... # split workload ... per_worker = int(math.ceil((self.end - self.start) / float(worker_info.num_workers))) ... worker_id = worker_info.id ... iter_start = self.start + worker_id * per_worker ... iter_end = min(iter_start + per_worker, self.end) ... return iter(range(iter_start, iter_end)) ... Example 2: splitting workload across all workers using for indices in batch_sampler:
01:> class MyIterableDataset(torch.utils.data.IterableDataset): ... def init(self, start, end): ... super(MyIterableDataset).init() ... assert end > start, "this example code only works with end >= start" ... self.start = start ... self.end = end ... ... def iter(self): ... return iter(range(self.start, self.end)) ... class torch.utils.data.TensorDataset(*tensors) Dataset wrapping tensors. Each sample will be retrieved by indexing tensors along the first dimension. Parameters *tensors () – tensors that have the same size of the first dimension. class torch.utils.data.StackDataset(*args, **kwargs) Dataset as a stacking of multiple datasets. This class is useful to assemble different parts of complex input data, given as datasets. Example > images = ImageDataset()texts = TextDataset()tuple_stack = StackDataset(images, texts) tuple_stack[0] == (images[0], texts[0]) dict_stack = StackDataset(image=images, text=texts) dict_stack[0] == {'image': images[0], 'text': texts[0]} Parameters
Dataset as a concatenation of multiple datasets. This class is useful to assemble different existing datasets. Parameters datasets (sequence) – List of datasets to be concatenated class torch.utils.data.ChainDataset(datasets) Dataset for chaining multiple s. This class is useful to assemble different existing dataset streams. The chaining operation is done on-the-fly, so concatenating large-scale datasets with this class will be efficient. Parameters datasets (iterable of ) – datasets to be chained together class torch.utils.data.Subset(dataset, indices) Subset of a dataset at specified indices. Parameters
General collate function that handles collection type of element within each batch. The function also opens function registry to deal with specific element types. default_collate_fn_map provides default collate functions for tensors, numpy arrays, numbers and strings. Parameters
Examples > def collate_tensor_fn(batch, *, collate_fn_map): Note Each collate function requires a positional argument for batch and a keyword argument for the dictionary of collate functions as collate_fn_map. torch.utils.data.default_collate(batch) Take in a batch of data and put the elements within the batch into a tensor with an additional outer dimension - batch size. The exact output type can be a , a Sequence of , a Collection of , or left unchanged, depending on the input type. This is used as the default function for collation when batch_size or batch_sampler is defined in . Here is the general input type (based on the type of the element within the batch) to output type mapping: -> (with an added outer dimension batch size) NumPy Arrays -> float -> int -> str -> str (unchanged) bytes -> bytes (unchanged) Mapping[K, V_i] -> Mapping[K, default_collate([V_1, V_2, …])] NamedTuple[V1_i, V2_i, …] -> NamedTuple[default_collate([V1_1, V1_2, …]), default_collate([V2_1, V2_2, …]), …] Parameters batch – a single batch to be collated Examples for indices in batch_sampler:
0torch.utils.data.default_convert(data) Convert each NumPy array element into a . If the input is a Sequence, Collection, or Mapping, it tries to convert each element inside to a . If the input is not an NumPy array, it is left unchanged. This is used as the default function for collation when both batch_sampler and batch_size are NOT defined in . The general input type to output type mapping is similar to that of . See the description there for more details. Parameters data – a single data point to be converted Examples for indices in batch_sampler:
1torch.utils.data.get_worker_info() Returns the information about the current iterator worker process. When called in a worker, this returns an object guaranteed to have the following attributes:
When called in the main process, this returns class SimpleCustomBatch: def collate_wrapper(batch): inps = torch.arange(10 5, dtype=torch.float32).view(10, 5)
tgts = torch.arange(10 5, dtype=torch.float32).view(10, 5)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset, batch_size=2, collate_fn=collate_wrapper, for batch_ndx, sample in enumerate(loader):
0.Note When used in a for indices in batch_sampler:
01 passed over to , this method can be useful to set up each worker process differently, for instance, using dataset_iter = iter(dataset) for indices in batch_sampler:
29 to configure the dataset_iter = iter(dataset) for indices in batch_sampler:
0 object to only read a specific fraction of a sharded dataset, or use dataset_iter = iter(dataset) for indices in batch_sampler:
23 to seed other libraries used in dataset code.Return type [WorkerInfo] torch.utils.data.random_split(dataset, lengths, generator=<torch._C.Generator object>) Randomly split a dataset into non-overlapping new datasets of given lengths. If a list of fractions that sum up to 1 is given, the lengths will be computed automatically as floor(frac * len(dataset)) for each fraction provided. After computing the lengths, if there are any remainders, 1 count will be distributed in round-robin fashion to the lengths until there are no remainders left. Optionally fix the generator for reproducible results, e.g.: Example for indices in batch_sampler:
2Parameters
[[T]] class torch.utils.data.Sampler(data_source\=None) Base class for all Samplers. Every Sampler subclass has to provide an dataset_iter = iter(dataset) for indices in batch_sampler:
7 method, providing a way to iterate over indices or lists of indices (batches) of dataset elements, and a dataset_iter = iter(dataset) for indices in batch_sampler:
2 method that returns the length of the returned iterators.Parameters data_source () – This argument is not used and will be removed in 2.2.0. You may still have custom implementation that utilizes it. Example for indices in batch_sampler:
3Note The dataset_iter = iter(dataset) for indices in batch_sampler:
2 method isn’t strictly required by , but is expected in any calculation involving the length of a .class torch.utils.data.SequentialSampler(data_source) Samples elements sequentially, always in the same order. Parameters data_source () – dataset to sample from class torch.utils.data.RandomSampler(data_source, replacement\=False, num_samples\=None, generator\=None) Samples elements randomly. If without replacement, then sample from a shuffled dataset. If with replacement, then user can specify dataset_iter = iter(dataset) for indices in batch_sampler:
37 to draw.Parameters
Samples elements randomly from a given list of indices, without replacement. Parameters
Samples elements from dataset_iter = iter(dataset) for indices in batch_sampler:
39 with given probabilities (weights).Parameters
Example for indices in batch_sampler:
4class torch.utils.data.BatchSampler(sampler, batch_size, drop_last) Wraps another sampler to yield a mini-batch of indices. Parameters
Example for indices in batch_sampler:
5class torch.utils.data.distributed.DistributedSampler(dataset, num_replicas\=None, rank\=None, shuffle\=True, seed\=0, drop_last\=False) Sampler that restricts data loading to a subset of the dataset. It is especially useful in conjunction with . In such a case, each process can pass a dataset_iter = iter(dataset) for indices in batch_sampler:
44 instance as a sampler, and load a subset of the original dataset that is exclusive to it.Note Dataset is assumed to be of constant size and that any instance of it always returns the same elements in the same order. Parameters
Warning In distributed mode, calling the dataset_iter = iter(dataset) for indices in batch_sampler:
55 method at the beginning of each epoch before creating the for indices in batch_sampler:
8 iterator is necessary to make shuffling work properly across multiple epochs. Otherwise, the same ordering will be always used. |