PyTorch and MindSpore API Mapping Table

Mapping between PyTorch APIs and MindSpore APIs, which is provided by the community. There may be differences in parameters, inputs, outputs, logic functions, and specific scenarios. For details, see the description of each API or the difference comparison provided.

More MindSpore developers are also welcome to participate in improving the mapping content. For more information on the differences in framework mechanisms of PyTorch and MindSpore, see: Optimizer Comparison, Random Number Strategy Comparison, and Parameter Initialization Comparison.

API Mapping Consistency Criteria and Exceptions

API mapping consistency criteria: consistent API function, consistent number or sequence of parameters, consistent parameter data type, consistent default value, consistent parameter name. Satisfying all the consistency conditions at the same time is considered as consistent API mapping.

The API mapping is also consistent in the following exception scenarios:

Exception Scenario 1: Compared to the API mapping consistency criteria, only the input data types of API parameters are not supported in the same range, including the following 3 sub-scenarios:

(1) MindSpore API supports passing parameters of int, float, bool, but does not support passing parameters of small bit-width data types such as int8 or float64.

(2) MindSpore API does not support passing parameters of plural type.

Exception Scenario 2: Compared to MindSpore APIss, the extra parameters of PyTorch API are general difference parameters. General difference parameters exist because PyTorch has some parameters that are added for non-functionality such as performance optimization, and the performance optimization mechanism of MindSpore is different from that of PyTorch.

Exception Scenario 3: If it can be guaranteed that MindSpore API uses the default configuration (or that the user does not configure it), MindSpore API can implement the same functionality as the PyTorch API, and MindSpore API has more parameters than PyTorch API. The functionality is not considered a difference.

Exception Scenario 4: MindSpore sets the default value of the parameters related to the PyTorch overloading mechanism in the API to None, and the corresponding parameters of the PyTorch counterpart API have no default value.

The following is an example of the exception scenario 4. In PyTorch 2.1, torch.argmax has two API overloads in the form of torch.argmax(input) and torch.argmax(input, dim, keepdim=False), respectively, where torch.argmax(input) returns the index of the maximum value element in the input Tensor, and torch.argmax(input, dim, keepdim=False) returns the index of the maximum value of the input Tensor on the specified axis.

mindspore.mint.argmax has only one API form, namely mindspore.mint.argmax(input, dim=None, keepdim=False), but mindspore.mint.argmax(input) and torch.argmax(input) have the same function. mindspore.mint.argmax(input, dim, keepdim) and torch.argmax(input, dim, keepdim)have the same function. Compared to torch.argmax, the default value of mindspore.ops.argmax parameter dim is set to None only to adapt the two API overload forms of torch.argmax, so the exception scenario 4 is also considered to be consistent API mapping.

General Difference Parameter Table

Because of the framework mechanism, MindSpore does not provide the following parameters for PyTorch:

Parameter Names	Functions	Descriptions
out	Indicates the output Tensor	Assign the operation result to the out parameter, not supported in MindSpore.
layout	Indicates the memory distribution strategy	PyTorch supports torch.striped and torch.split_coo, not supported in MindSpore.
device	Indicates the Tensor storage location	Including device type and optional device number, MindSpore currently supports operator or network-level device scheduling.
requires_grad	Indicates whether to update the gradient	MindSpore can be accessed through the Parameter.requires_grad attribute to control.
pin_memory	Indicates whether to use locking page memory	Not supported in MindSpore.
memory_format	Indicates the memory format of the Tensor	Not supported in MindSpore.
stable	Indicates whether the sorting is stable	Generally used in the API of Sorting algorithm, not supported in MindSpore.
sparse_grad	Indicates whether to perform sparsification on the gradient	Not supported in MindSpore.
size_average	The deprecated parameter in PyTorch	The reduction parameter can be used instead.
reduce	The deprecated parameter in PyTorch	The reduction parameter can be used instead.

torch

PyTorch 2.1 APIs	MindSpore APIs	Descriptions
torch.abs	mindspore.mint.abs	Consistent
torch.acos	mindspore.mint.acos	Consistent
torch.acosh	mindspore.mint.acosh	Consistent
torch.add	mindspore.mint.add	Consistent
torch.addbmm	mindspore.mint.addbmm	Consistent
torch.addmm	mindspore.mint.addmm	Consistent
torch.addmm	mindspore.mint.addmv	Consistent
torch.all	mindspore.mint.all	Consistent
torch.allclose	mindspore.mint.allclose	Consistent
torch.amax	mindspore.mint.amax	The functions are consistent, but the default value of dim is different.
torch.amin	mindspore.mint.amin	The functions are consistent, but the default value of dim is different.
torch.any	mindspore.mint.any	Consistent
torch.arange	mindspore.mint.arange	The functions are consistent, but the default value of end is different.
torch.arccos	mindspore.mint.arccos	Consistent
torch.arccosh	mindspore.mint.arccosh	Consistent
torch.arcsin	mindspore.mint.arcsin	Consistent functions, inconsistent parameter names.
torch.arcsinh	mindspore.mint.arcsinh	Consistent
torch.arctan	mindspore.mint.arctan	Consistent
torch.arctan2	mindspore.mint.arctan2	Consistent
torch.arctanh	mindspore.mint.arctanh	Consistent
torch.argmax	mindspore.mint.argmax	Consistent
torch.argmin	mindspore.mint.argmin	Consistent
torch.argsort	mindspore.mint.argsort	Consistent
torch.asin	mindspore.mint.asin	Consistent
torch.asinh	mindspore.mint.asinh	Consistent
torch.atan	mindspore.mint.atan	Consistent
torch.atan2	mindspore.mint.atan2	Consistent
torch.atanh	mindspore.mint.atanh	Consistent
torch.baddbmm	mindspore.mint.baddbmm	Consistent
torch.bernoulli	mindspore.mint.bernoulli	Consistent
torch.bincount	mindspore.mint.bincount	Consistent
torch.bitwise_and	mindspore.mint.bitwise_and	Consistent
torch.bitwise_or	mindspore.mint.bitwise_or	Consistent
torch.bitwise_xor	mindspore.mint.bitwise_xor	Consistent
torch.bmm	mindspore.mint.bmm	Consistent
torch.broadcast_to	mindspore.mint.broadcast_to	Consistent
torch.cat	mindspore.mint.cat	Consistent
torch.cdist	mindspore.mint.cdist	Consistent
torch.chunk	mindspore.mint.chunk	Consistent
torch.ceil	mindspore.mint.ceil	Consistent
torch.clamp	mindspore.mint.clamp	Consistent
torch.clone	mindspore.mint.clone	Consistent
torch.concat	mindspore.mint.concat	Consistent
torch.count_nonzero	mindspore.mint.count_nonzero	Consistent
torch.cos	mindspore.mint.cos	Consistent
torch.cosh	mindspore.mint.cosh	Consistent
torch.cross	mindspore.mint.cross	Consistent
torch.cummax	mindspore.mint.cummax	Consistent
torch.cummin	mindspore.mint.cummin	Consistent
torch.cumprod	mindspore.mint.cumprod	Consistent
torch.cumsum	mindspore.mint.cumsum	Consistent
torch.diff	mindspore.mint.diff	Consistent
torch.div	mindspore.mint.div	Consistent
torch.divide	mindspore.mint.divide	Consistent
torch.dot	mindspore.mint.dot	Consistent
torch.einsum	mindspore.mint.einsum	Consistent
torch.empty	mindspore.mint.empty	Consistent
torch.empty_like	mindspore.mint.empty_like	Consistent
torch.equal	mindspore.mint.equal	Consistent
torch.eq	mindspore.mint.eq	Consistent
torch.erf	mindspore.mint.erf	Consistent
torch.erfc	mindspore.mint.erfc	Consistent
torch.erfinv	mindspore.mint.erfinv	Consistent
torch.exp	mindspore.mint.exp	Consistent
torch.exp2	mindspore.mint.exp2	Consistent
torch.expm1	mindspore.mint.expm1	Consistent
torch.eye	mindspore.mint.eye	Consistent
torch.float_power	mindspore.mint.float_power	Consistent
torch.flatten	mindspore.mint.flatten	Consistent
torch.flip	mindspore.mint.flip	Consistent
torch.floor	mindspore.mint.floor	Consistent
torch.fmod	mindspore.mint.fmod	Consistent
torch.frac	mindspore.mint.frac	Consistent
torch.full	mindspore.mint.full	Consistent
torch.full_like	mindspore.mint.full_like	Consistent
torch.gather	mindspore.mint.gather	Consistent
torch.greater	mindspore.mint.greater	Consistent
torch.greater_equal	mindspore.mint.greater_equal	Consistent
torch.gt	mindspore.mint.gt	Consistent
torch.histc	mindspore.mint.histc	Consistent
torch.index_select	mindspore.mint.index_select	Consistent
torch.inverse	mindspore.mint.inverse	Consistent
torch.isclose	mindspore.mint.isclose	Consistent
torch.isfinite	mindspore.mint.isfinite	Consistent
torch.isinf	mindspore.mint.isinf	Consistent
torch.isneginf	mindspore.mint.isneginf	Consistent
torch.le	mindspore.mint.le	Consistent
torch.lerp	mindspore.mint.lerp	Consistent
torch.less	mindspore.mint.less	Consistent
torch.less_equal	mindspore.mint.less_equal	Consistent
torch.linspace	mindspore.mint.linspace	Consistent
torch.log	mindspore.mint.log	Consistent
torch.log2	mindspore.mint.log2	Consistent
torch.log10	mindspore.mint.log10	Consistent
torch.logaddexp	mindspore.mint.logaddexp	Consistent
torch.logsumexp	mindspore.mint.logsumexp	Consistent
torch.log1p	mindspore.mint.log1p	Consistent
torch.logical_and	mindspore.mint.logical_and	Consistent
torch.logical_not	mindspore.mint.logical_not	Consistent
torch.logical_or	mindspore.mint.logical_or	Consistent
torch.logical_xor	mindspore.mint.logical_xor	Consistent
torch.lt	mindspore.mint.lt	Consistent
torch.masked_select	mindspore.mint.masked_select	Consistent
torch.matmul	mindspore.mint.matmul	Consistent
torch.max	mindspore.mint.max	Consistent
torch.maximum	mindspore.mint.maximum	Consistent
torch.mean	mindspore.mint.mean	Consistent
torch.median	mindspore.mint.median	The functions are consistent, but the default value of dim is different.
torch.meshgrid	mindspore.mint.meshgrid	The functions are consistent, but the default value of indexing is different.
torch.mul	mindspore.mint.mul	Consistent
torch.min	mindspore.mint.min	Consistent
torch.minimum	mindspore.mint.minimum	Consistent
torch.mm	mindspore.mint.mm	Consistent
torch.multinomial	mindspore.mint.multinomial	Consistent
torch.mv	mindspore.mint.mv	Consistent
torch.nansum	mindspore.mint.nansum	The functions are consistent, but the default value of end is different.
torch.nan_to_num	mindspore.mint.nan_to_num	Consistent
torch.narrow	mindspore.mint.narrow	Consistent
torch.ne	mindspore.mint.ne	Consistent
torch.neg	mindspore.mint.neg	Consistent
torch.negative	mindspore.mint.negative	Consistent
torch.nonzero	mindspore.mint.nonzero	Consistent
torch.normal	mindspore.mint.normal	The parameters of interface overloading are different.
torch.norm	mindspore.mint.norm	Consistent
torch.ones	mindspore.mint.ones	Consistent
torch.ones_like	mindspore.mint.ones_like	Consistent
torch.outer	mindspore.mint.outer	Consistent
torch.permute	mindspore.mint.permute	Consistent
torch.polar	mindspore.mint.polar	Consistent
torch.pow	mindspore.mint.pow	Consistent
torch.prod	mindspore.mint.prod	Consistent
torch.rand	mindspore.mint.rand	Consistent
torch.rand_like	mindspore.mint.rand_like	Consistent
torch.randint	mindspore.mint.randint	Consistent
torch.randint_like	mindspore.mint.randint_like	The functions are consistent, but the default value of low is different.
torch.randn	mindspore.mint.randn	Consistent
torch.randn_like	mindspore.mint.randn_like	Consistent
torch.randperm	mindspore.mint.randperm	Consistent
torch.ravel	mindspore.mint.ravel	Consistent
torch.reciprocal	mindspore.mint.reciprocal	Consistent
torch.remainder	mindspore.mint.remainder	Consistent
torch.repeat_interleave	mindspore.mint.repeat_interleave	Consistent functions, PyTorch involves overloading.
torch.reshape	mindspore.mint.reshape	Consistent
torch.roll	mindspore.mint.roll	Consistent
torch.round	mindspore.mint.round	Consistent
torch.rsqrt	mindspore.mint.rsqrt	Consistent
torch.scatter	mindspore.mint.scatter	Consistent
torch.scatter_add	mindspore.mint.scatter_add	Consistent
torch.searchsorted	mindspore.mint.searchsorted	The functions are consistent, but the default value of side is different.
torch.select	mindspore.mint.select	Consistent
torch.sigmoid	mindspore.mint.sigmoid	Consistent
torch.sign	mindspore.mint.sign	Consistent
torch.sin	mindspore.mint.sin	Consistent
torch.sinc	mindspore.mint.sinc	Consistent
torch.sinh	mindspore.mint.sinh	Consistent
torch.sort	mindspore.mint.sort	Consistent
torch.split	mindspore.mint.split	Consistent
torch.sqrt	mindspore.mint.sqrt	Consistent
torch.square	mindspore.mint.square	Consistent
torch.squeeze	mindspore.mint.squeeze	Consistent
torch.stack	mindspore.mint.stack	Consistent
torch.std	mindspore.mint.std	Consistent
torch.std_mean	mindspore.mint.std_mean	Consistent
torch.sub	mindspore.mint.sub	Consistent
torch.sum	mindspore.mint.sum	Consistent
torch.swapaxes	mindspore.mint.swapaxes	Consistent
torch.t	mindspore.mint.t	Consistent
torch.tan	mindspore.mint.tan	Consistent
torch.tanh	mindspore.mint.tanh	Consistent
torch.tile	mindspore.mint.tile	Consistent
torch.topk	mindspore.mint.topk	The functions are consistent, but the default value of dim is different.
torch.trace	mindspore.mint.trace	Consistent
torch.transpose	mindspore.mint.transpose	Consistent
torch.tril	mindspore.mint.tril	Consistent
torch.triu	mindspore.mint.triu	Consistent
torch.trunc	mindspore.mint.trunc	Consistent
torch.unbind	mindspore.mint.unbind	Consistent
torch.unique	mindspore.mint.unique	Consistent
torch.unique_consecutive	mindspore.mint.unique_consecutive	Consistent
torch.unsqueeze	mindspore.mint.unsqueeze	Consistent
torch.var	mindspore.mint.var	Consistent
torch.var_mean	mindspore.mint.var_mean	Consistent
torch.where	mindspore.mint.where	Consistent
torch.xlogy	mindspore.mint.xlogy	Consistent
torch.zeros	mindspore.mint.zeros	Consistent
torch.zeros_like	mindspore.mint.zeros_like	Consistent

torch.linalg

PyTorch 2.1 APIs	MindSpore APIs	说明
torch.linalg.inv	mindspore.mint.linalg.inv	Consistent
torch.linalg.matrix_norm	mindspore.mint.linalg.matrix_norm	Consistent
torch.linalg.norm	mindspore.mint.linalg.norm	Consistent
torch.linalg.vector_norm	mindspore.mint.linalg.vector_norm	Consistent

torch.distributed

PyTorch 2.1 APIs	MindSpore APIs	Descriptions
torch.distributed.P2POp	mindspore.mint.distributed.P2POp	Consistent
torch.distributed.all_gather	mindspore.mint.distributed.all_gather	Consistent
torch.distributed.all_gather_into_tensor	mindspore.mint.distributed.all_gather_into_tensor	Consistent
torch.distributed.all_gather_object	mindspore.mint.distributed.all_gather_object	Consistent
torch.distributed.all_reduce	mindspore.mint.distributed.all_reduce	Consistent
torch.distributed.all_to_all_single	mindspore.mint.distributed.all_to_all_single	Consistent
torch.distributed.all_to_all	mindspore.mint.distributed.all_to_all	Consistent
torch.distributed.barrier	mindspore.mint.distributed.barrier	Consistent
torch.distributed.batch_isend_irecv	mindspore.mint.distributed.batch_isend_irecv	Consistent
torch.distributed.broadcast	mindspore.mint.distributed.broadcast	Consistent
torch.distributed.distributed.broadcast_object_list	mindspore.mint.distributed.distributed.broadcast_object_list	Consistent
	mindspore.mint.distributed.destroy_process_group	Unique to MindSpore
torch.distributed.gather	mindspore.mint.distributed.gather	Consistent
torch.distributed.gather_object	mindspore.mint.distributed.gather	Consistent
torch.distributed.get_backend	mindspore.mint.distributed.get_backend	Consistent
torch.distributed.get_global_rank	mindspore.mint.distributed.get_global_rank	Consistent
torch.distributed.get_group_rank	mindspore.mint.distributed.get_group_rank	Consistent
torch.distributed.get_process_group_ranks	mindspore.mint.distributed.get_process_group_ranks	Consistent
torch.distributed.get_rank	mindspore.mint.distributed.get_rank	Consistent
torch.distributed.get_world_size	mindspore.mint.distributed.get_world_size	Consistent
torch.distributed.init_process_group	mindspore.mint.distributed.init_process_group	Consistent functions, inconsistent parameter names.
torch.distributed.irecv	mindspore.mint.distributed.irecv	Consistent
torch.distributed.isend	mindspore.mint.distributed.isend	Consistent
torch.distributed.new_group	mindspore.mint.distributed.new_group	Consistent functions, MindSpore has an additional parameter group_desc = None.
torch.distributed.recv	mindspore.mint.distributed.recv	Consistent
torch.distributed.reduce	mindspore.mint.distributed.reduce	Consistent
torch.distributed.reduce_scatter	mindspore.mint.distributed.reduce_scatter	Consistent
torch.distributed.reduce_scatter_tensor	mindspore.mint.distributed.reduce_scatter_tensor	Consistent
torch.distributed.scatter	mindspore.mint.distributed.scatter	The functions are consistent, but the default value of scatter_list is different.
torch.distributed.scatter_object_list	mindspore.mint.distributed.scatter_object_list	Consistent
torch.distributed.send	mindspore.mint.distributed.send	Consistent

torch.nn

PyTorch 2.1 APIs	MindSpore APIs	Descriptions
torch.nn.AdaptiveAvgPool1d	mindspore.mint.nn.AdaptiveAvgPool1d	Consistent
torch.nn.AdaptiveAvgPool2d	mindspore.mint.nn.AdaptiveAvgPool2d	Consistent
torch.nn.AdaptiveAvgPool3d	mindspore.mint.nn.AdaptiveAvgPool3d	Consistent
torch.nn.AvgPool2d	mindspore.mint.nn.AvgPool2d	Consistent
torch.nn.BCELoss	mindspore.mint.nn.BCELoss	Consistent
torch.nn.BCEWithLogitsLoss	mindspore.mint.nn.BCEWithLogitsLoss	Consistent
torch.nn.BatchNorm1d	mindspore.mint.nn.BatchNorm1d	Consistent
torch.nn.BatchNorm2d	mindspore.mint.nn.BatchNorm2d	Consistent
torch.nn.BatchNorm3d	mindspore.mint.nn.BatchNorm3d	Consistent
torch.nn.ConstantPad1d	mindspore.mint.nn.ConstantPad1d	Consistent
torch.nn.ConstantPad2d	mindspore.mint.nn.ConstantPad2d	Consistent
torch.nn.ConstantPad3d	mindspore.mint.nn.ConstantPad3d	Consistent
torch.nn.Conv2d	mindspore.mint.nn.Conv2d	Consistent
torch.nn.Conv3d	mindspore.mint.nn.Conv3d	Consistent
torch.nn.ConvTranspose2d	mindspore.mint.nn.ConvTranspose2d	Consistent
torch.nn.CrossEntropyLoss	mindspore.mint.nn.CrossEntropyLoss	Consistent
torch.nn.Dropout	mindspore.mint.nn.Dropout	Consistent
torch.nn.Dropout2d	mindspore.mint.nn.Dropout2d	Consistent functions, MindSpore has no parameter inplace.
torch.nn.ELU	mindspore.mint.nn.ELU	Consistent functions, MindSpore has no parameter inplace.
torch.nn.Embedding	mindspore.mint.nn.Embedding	Consistent
torch.nn.Flatten	mindspore.mint.flatten	Consistent
torch.nn.Fold	mindspore.mint.nn.Fold	Consistent
torch.nn.GroupNorm	mindspore.mint.nn.Fold	Consistent
torch.nn.GELU	mindspore.mint.nn.GELU	Consistent
torch.nn.GroupNorm	mindspore.mint.nn.GroupNorm	Consistent
torch.nn.Hardshrink	mindspore.mint.nn.Hardshrink	Consistent
torch.nn.Hardsigmoid	mindspore.mint.nn.Hardsigmoid	Consistent
torch.nn.Hardswish	mindspore.mint.nn.Hardswish	Consistent
torch.nn.Identity	mindspore.mint.nn.Identity	Consistent
torch.nn.L1Loss	mindspore.mint.nn.L1Loss	Consistent
torch.nn.LayerNorm	mindspore.mint.nn.LayerNorm	Consistent
torch.nn.Linear	mindspore.mint.nn.Linear	Consistent function, MindSpore has two different parameters: weight_init = None and bias_init = None.
torch.nn.LogSigmoid	mindspore.mint.nn.LogSigmoid	Consistent
torch.nn.LogSoftMax	mindspore.mint.nn.LogSoftmax	Consistent
torch.nn.MSELoss	mindspore.mint.nn.MSELoss	Consistent
torch.nn.MaxUnpool2d	mindspore.mint.nn.MaxUnpool2d	Consistent
torch.nn.Mish	mindspore.mint.nn.Mish	Consistent
torch.nn.NLLLoss	mindspore.mint.nn.NLLLoss	Consistent
torch.nn.PReLU	mindspore.mint.nn.PReLU	Consistent
torch.nn.ReLU	mindspore.mint.nn.ReLU	Consistent
torch.nn.ReLU6	mindspore.mint.nn.ReLU6	Consistent
torch.nn.ReflectionPad1d	mindspore.mint.nn.ReflectionPad1d	Consistent
torch.nn.ReflectionPad2d	mindspore.mint.nn.ReflectionPad2d	Consistent
torch.nn.ReflectionPad3d	mindspore.mint.nn.ReflectionPad3d	Consistent
torch.nn.ReplicationPad1d	mindspore.mint.nn.ReplicationPad1d	Consistent
torch.nn.ReplicationPad2d	mindspore.mint.nn.ReplicationPad2d	Consistent
torch.nn.ReplicationPad3d	mindspore.mint.nn.ReplicationPad3d	Consistent
torch.nn.SeLU	mindspore.mint.nn.SELU	Consistent
torch.nn.SiLU	mindspore.mint.nn.SiLU	Consistent
torch.nn.SmoothL1Loss	mindspore.mint.nn.SmoothL1Loss	Consistent
torch.nn.Softmax	mindspore.mint.nn.Softmax	Consistent
torch.nn.Softshrink	mindspore.mint.nn.Softshrink	Consistent
torch.nn.SyncBatchNorm	mindspore.mint.nn.SyncBatchNorm	Consistent
torch.nn.Tanh	mindspore.mint.nn.Tanh	Consistent
torch.nn.Unfold	mindspore.mint.nn.Unfold	Consistent
torch.nn.Upsample	mindspore.mint.nn.Upsample	Consistent
torch.nn.ZeroPad1d	mindspore.mint.nn.ZeroPad1d	Consistent
torch.nn.ZeroPad2d	mindspore.mint.nn.ZeroPad2d	Consistent
torch.nn.ZeroPad3d	mindspore.mint.nn.ZeroPad3d	Consistent

torch.nn.functional

PyTorch 2.1 APIs	MindSpore APIs	Descriptions
torch.nn.functional.adaptive_avg_pool1d	mindspore.mint.nn.functional.adaptive_avg_pool1d	Consistent
torch.nn.functional.adaptive_avg_pool2d	mindspore.mint.nn.functional.adaptive_avg_pool2d	Consistent
torch.nn.functional.avg_pool1d	mindspore.mint.nn.functional.avg_pool1d	Consistent
torch.nn.functional.avg_pool2d	mindspore.mint.nn.functional.avg_pool2d	Consistent
torch.nn.functional.batch_norm	mindspore.mint.nn.functional.batch_norm	Consistent
torch.nn.functional.binary_cross_entropy	mindspore.mint.nn.functional.binary_cross_entropy	Consistent
torch.nn.functional.binary_cross_entropy_with_logits	mindspore.mint.nn.functional.binary_cross_entropy_with_logits	Consistent
torch.nn.functional.conv2d	mindspore.mint.nn.functional.conv2d	Consistent
torch.nn.functional.conv3d	mindspore.mint.nn.functional.conv3d	Consistent
torch.nn.functional.conv_transpose2d	mindspore.mint.nn.functional.conv_transpose2d	Consistent
torch.nn.functional.dropout	mindspore.mint.nn.functional.dropout	Consistent
torch.nn.functional.dropout2d	mindspore.mint.nn.functional.dropout2d	Consistent functions, MindSpore does not contain the parameter inplace.
torch.nn.functional.elu	mindspore.mint.nn.functional.elu	Consistent functions, MindSpore does not contain the parameter inplace.
torch.nn.functional.embedding	mindspore.mint.nn.functional.embedding	Consistent functions, MindSpore does not contain the parameter sparse.
torch.nn.functional.fold	mindspore.mint.nn.functional.fold	Consistent
torch.nn.functional.gelu	mindspore.mint.nn.functional.gelu	Consistent
torch.nn.functional.grid_sample	mindspore.mint.nn.functional.grid_sample	Consistent functions, but the default value of align_corners is different.
torch.nn.functional.hardshrink	mindspore.mint.nn.functional.hardshrink	Consistent
torch.nn.functional.hardsigmoid	mindspore.mint.nn.functional.hardsigmoid	Consistent functions, MindSpore has no parameter inplace.
torch.nn.functional.hardswish	mindspore.mint.nn.functional.hardswish	Consistent functions, MindSpore has no parameter inplace.
torch.nn.functional.interpolate	mindspore.mint.nn.functional.interpolate	Consistent functions, MindSpore has no parameter antialias.
torch.nn.functional.l1_loss	mindspore.mint.nn.functional.l1_loss	Consistent](https://www.mindspore.cn/docs/en/master/note/api_mapping/pytorch_api_mapping.html#api-mapping-consistency-criteria-and-exceptions)
torch.nn.functional.leaky_relu	mindspore.mint.nn.functional.leaky_relu	Consistent functions, MindSpore has no parameter inplace.
torch.nn.functional.linear	mindspore.mint.nn.functional.linear	Consistent
torch.nn.functional.log_softmax	mindspore.mint.nn.functional.log_softmax	Consistent
torch.nn.functional.logsigmoid	mindspore.mint.nn.functional.logsigmoid	Consistent
torch.nn.functional.max_pool2d	mindspore.mint.nn.functional.max_pool2d	Consistent
torch.nn.functional.max_unpool2d	mindspore.mint.nn.functional.max_unpool2d	Consistent
torch.nn.functional.mish	mindspore.mint.nn.functional.mish	Consistent functions, MindSpore has no parameter inplace.
torch.nn.functional.mse_loss	mindspore.mint.nn.functional.mse_loss	Consistent
torch.nn.functional.nll_loss	mindspore.mint.nn.functional.nll_loss	Consistent
torch.nn.functional.normalize	mindspore.mint.nn.functional.normalize	Consistent
torch.nn.functional.one_hot	mindspore.mint.nn.functional.one_hot	Consistent
torch.nn.functional.pad	mindspore.mint.nn.functional.pad	Consistent
torch.nn.functional.prelu	mindspore.mint.nn.functional.prelu	Consistent
torch.nn.functional.relu	mindspore.mint.nn.functional.relu	Consistent
torch.nn.functional.relu6	mindspore.mint.nn.functional.relu6	Consistent
torch.nn.functional.relu_	mindspore.mint.nn.functional.relu_	Consistent
torch.nn.functional.selu	mindspore.mint.nn.functional.selu	Consistent functions, MindSpore does not contain parameter inplace.
torch.nn.functional.sigmoid	mindspore.mint.nn.functional.sigmoid	Consistent
torch.nn.functional.silu	mindspore.mint.nn.functional.silu	Consistent functions, MindSpore does not contain parameter inplace.
torch.nn.functional.smooth_l1_loss	mindspore.mint.nn.functional.smooth_l1_loss	Consistent
torch.nn.functional.softmax	mindspore.mint.nn.functional.softmax	Consistent
torch.nn.functional.softplus	mindspore.mint.nn.functional.softplus	Consistent
torch.nn.functional.softshrink	mindspore.mint.nn.functional.softshrink	Consistent
torch.nn.functional.tanh	mindspore.mint.nn.functional.tanh	Consistent
torch.nn.functional.unfold	mindspore.mint.nn.functional.unfold	Consistent

torch.special

PyTorch 2.1 APIs	MindSpore APIs	说明
torch.special.erfc	mindspore.mint.special.erfc	Consistent
torch.special.exp2	mindspore.mint.special.exp2	Consistent
torch.special.expm1	mindspore.mint.special.expm1	Consistent
torch.special.log1p	mindspore.mint.special.log1p	Consistent
torch.special.log_softmax	mindspore.mint.special.log_softmax	Consistent
torch.special.round	mindspore.mint.special.round	Consistent
torch.special.sinc	mindspore.mint.special.sinc	Consistent

torch.Tensor

PyTorch 2.1 APIs	MindSpore APIs	Descriptions
torch.Tensor.abs	mindspore.Tensor.abs	Consistent
torch.Tensor.absolute	mindspore.Tensor.absolute	Consistent
torch.Tensor.add	mindspore.Tensor.add	Consistent
torch.Tensor.all	mindspore.Tensor.all	Consistent
torch.Tensor.any	mindspore.Tensor.any	Consistent
torch.Tensor.argmax	mindspore.Tensor.argmax	Consistent
torch.Tensor.arctan2	mindspore.Tensor.arctan2	Consistent
torch.Tensor.argmin	mindspore.Tensor.argmin	Consistent
torch.Tensor.atan2	mindspore.Tensor.atan2	Consistent
torch.Tensor.bool	mindspore.Tensor.bool	Consistent
torch.Tensor.ceil	mindspore.Tensor.ceil	Consistent
torch.Tensor.chunk	mindspore.Tensor.chunk	Consistent
torch.Tensor.clamp	mindspore.Tensor.clamp	Consistent
torch.Tensor.clamp_	mindspore.Tensor.clamp_	Consistent
torch.Tensor.clip	mindspore.Tensor.clip	Consistent
torch.Tensor.cos	mindspore.Tensor.cos	Consistent
torch.Tensor.cumsum	mindspore.Tensor.cumsum	Consistent
torch.Tensor.div	mindspore.Tensor.div	Consistent
torch.Tensor.divide	mindspore.Tensor.divide	Consistent
torch.Tensor.eq	mindspore.Tensor.eq	Consistent
torch.Tensor.erf	mindspore.Tensor.erf	Consistent
torch.Tensor.exp	mindspore.Tensor.exp	Consistent
torch.Tensor.expand_as	mindspore.Tensor.expand_as	Consistent
torch.Tensor.flatten	mindspore.Tensor.flatten	Consistent
torch.Tensor.float	mindspore.Tensor.float	Consistent
torch.Tensor.floor	mindspore.Tensor.floor	Consistent
torch.Tensor.gather	mindspore.Tensor.gather	Consistent
torch.Tensor.greater	mindspore.Tensor.greater	Consistent
torch.Tensor.gt	mindspore.Tensor.gt	Consistent
torch.Tensor.half	mindspore.Tensor.half	Consistent
torch.Tensor.index_select	mindspore.Tensor.index_select	Consistent
torch.Tensor.int	mindspore.Tensor.int	Consistent
torch.Tensor.isfinite	mindspore.Tensor.isfinite	Consistent
torch.Tensor.isnan	mindspore.Tensor.isnan	Consistent
torch.Tensor.item	mindspore.Tensor.item	Consistent
torch.Tensor.le	mindspore.Tensor.le	Consistent
torch.Tensor.less	mindspore.Tensor.less	Consistent
torch.Tensor.less_equal	mindspore.Tensor.less_equal	Consistent
torch.Tensor.log	mindspore.Tensor.log	Consistent
torch.Tensor.logical_and	mindspore.Tensor.logical_and	Consistent
torch.Tensor.logical_not	mindspore.Tensor.logical_not	Consistent
torch.Tensor.logical_or	mindspore.Tensor.logical_or	Consistent
torch.Tensor.long	mindspore.Tensor.long	Consistent
torch.Tensor.lt	mindspore.Tensor.lt	Consistent
torch.Tensor.masked_fill	mindspore.Tensor.masked_fill	Consistent
torch.Tensor.masked_select	mindspore.Tensor.masked_select	Consistent
torch.Tensor.matmul	mindspore.Tensor.matmul	Consistent
torch.Tensor.max	mindspore.Tensor.max	Consistent
torch.Tensor.maximum	mindspore.Tensor.maximum	Consistent
torch.Tensor.mean	mindspore.Tensor.mean	Consistent
torch.Tensor.min	mindspore.Tensor.min	Consistent
torch.Tensor.minimum	mindspore.Tensor.minimum	Consistent
torch.Tensor.mul	mindspore.Tensor.mul	Consistent
torch.Tensor.nan_to_num	mindspore.Tensor.nan_to_num	Consistent
torch.Tensor.ne	mindspore.Tensor.ne	Consistent
torch.Tensor.neg	mindspore.Tensor.neg	Consistent
torch.Tensor.negative	mindspore.Tensor.negative	Consistent
torch.Tensor.not_equal	mindspore.Tensor.not_equal	Consistent
torch.Tensor.pow	mindspore.Tensor.pow	Consistent
torch.Tensor.prod	mindspore.Tensor.prod	Consistent
torch.Tensor.reciprocal	mindspore.Tensor.reciprocal	Consistent
torch.Tensor.remainder	mindspore.Tensor.remainder	Consistent
torch.Tensor.repeat_interleave	mindspore.Tensor.repeat_interleave	Consistent
torch.Tensor.reshape	mindspore.Tensor.reshape	Consistent
torch.Tensor.round	mindspore.Tensor.round	Consistent
torch.Tensor.rsqrt	mindspore.Tensor.rsqrt	Consistent
torch.Tensor.scatter	mindspore.Tensor.scatter	Consistent
torch.Tensor.scatter_add	mindspore.Tensor.scatter_add	Consistent
torch.Tensor.sigmoid	mindspore.Tensor.sigmoid	Consistent
torch.Tensor.sin	mindspore.Tensor.sin	Consistent
torch.Tensor.sort	mindspore.Tensor.sort	Consistent
torch.Tensor.split	mindspore.Tensor.split	Consistent
torch.Tensor.sqrt	mindspore.Tensor.sqrt	Consistent
torch.Tensor.square	mindspore.Tensor.square	Consistent
torch.Tensor.sub	mindspore.Tensor.sub	Consistent
torch.Tensor.sum	mindspore.Tensor.sum	Consistent
torch.Tensor.t	mindspore.Tensor.t	Consistent
torch.Tensor.tanh	mindspore.Tensor.tanh	Consistent
torch.Tensor.tile	mindspore.Tensor.tile	Consistent
torch.Tensor.topk	mindspore.Tensor.topk	Consistent
torch.Tensor.tril	mindspore.Tensor.tril	Consistent
torch.Tensor.trunc	mindspore.Tensor.trunc	Consistent
torch.Tensor.view_as	mindspore.Tensor.view_as	Consistent
torch.Tensor.where	mindspore.Tensor.where	Consistent

torch.optim

PyTorch 2.1 APIs	MindSpore APIs	Descriptions
torch.optim.Adam	mindspore.mint.optim.Adam	The functions are consistent, but PyTorch has some optimization parameters
torch.optim.AdamW	mindspore.mint.optim.AdamW	The functions are consistent, but PyTorch has some optimization parameters

torch.utils

PyTorch 1.8.1 APIs	MindSpore APIs	Descriptions
torch.utils.data.DataLoader	mindspore.dataset.GeneratorDataset	diff
torch.utils.data.distributed.DistributedSampler	mindspore.dataset.DistributedSampler	diff
torch.utils.data.RandomSampler	mindspore.dataset.RandomSampler	diff
torch.utils.data.SequentialSampler	mindspore.dataset.SequentialSampler	diff
torch.utils.data.SubsetRandomSampler	mindspore.dataset.SubsetRandomSampler	diff
torch.utils.data.WeightedRandomSampler	mindspore.dataset.WeightedRandomSampler	diff
torch.utils.checkpoint.checkpoint	mindspore.nn.Cell.recompute	diff

torchaudio

The current API mapping table corresponds to PyTorch version 1.8.1, and Python version is no higher than Python 3.9.

TorchAudio 0.8.1 APIs	MindSpore APIs	Descriptions
torchaudio.datasets.CMUARCTIC	mindspore.dataset.CMUArcticDataset	diff
torchaudio.datasets.GTZAN	mindspore.dataset.GTZANDataset	diff
torchaudio.datasets.LIBRITTS	mindspore.dataset.LibriTTSDataset	diff
torchaudio.datasets.LJSPEECH	mindspore.dataset.LJSpeechDataset	diff
torchaudio.datasets.SPEECHCOMMANDS	mindspore.dataset.SpeechCommandsDataset	diff
torchaudio.datasets.TEDLIUM	mindspore.dataset.TedliumDataset	diff
torchaudio.datasets.YESNO	mindspore.dataset.YesNoDataset	diff
torchaudio.transforms.AmplitudeToDB	mindspore.dataset.audio.AmplitudeToDB	diff
torchaudio.transforms.ComplexNorm	mindspore.dataset.audio.ComplexNorm	Consistent
torchaudio.transforms.ComputeDeltas	mindspore.dataset.audio.ComputeDeltas	The functions are consistent, but the parameter names are inconsistent.
torchaudio.transforms.Fade	mindspore.dataset.audio.Fade	Consistent
torchaudio.transforms.FrequencyMasking	mindspore.dataset.audio.FrequencyMasking	diff
torchaudio.transforms.GriffinLim	mindspore.dataset.audio.GriffinLim	diff
torchaudio.transforms.InverseMelScale	mindspore.dataset.audio.InverseMelScale	diff
torchaudio.transforms.MelScale	mindspore.dataset.audio.MelScale	diff
torchaudio.transforms.MelSpectrogram	mindspore.dataset.audio.MelSpectrogram	diff
torchaudio.transforms.MFCC	mindspore.dataset.audio.MFCC	Consistent
torchaudio.transforms.MuLawEncoding	mindspore.dataset.audio.MuLawEncoding	Consistent
torchaudio.transforms.MuLawDecoding	mindspore.dataset.audio.MuLawDecoding	Consistent
torchaudio.transforms.Resample	mindspore.dataset.audio.Resample	diff
torchaudio.transforms.SlidingWindowCmn	mindspore.dataset.audio.SlidingWindowCmn	Consistent
torchaudio.transforms.SpectralCentroid	mindspore.dataset.audio.SpectralCentroid	diff
torchaudio.transforms.Spectrogram	mindspore.dataset.audio.Spectrogram	diff
torchaudio.transforms.TimeMasking	mindspore.dataset.audio.TimeMasking	diff
torchaudio.transforms.TimeStretch	mindspore.dataset.audio.TimeStretch	Consistent
torchaudio.transforms.Vad	mindspore.dataset.audio.Vad	Consistent
torchaudio.transforms.Vol	mindspore.dataset.audio.Vol	Consistent

torchtext

The current API mapping table corresponds to PyTorch version 1.8.1, and Python version is no higher than Python 3.9.

TorchText 0.9.1 APIs	MindSpore APIs	Descriptions
torchtext.data.functional.custom_replace	mindspore.dataset.text.RegexReplace	diff
torchtext.data.functional.load_sp_model	mindspore.dataset.text.SentencePieceTokenizer	diff
torchtext.data.functional.numericalize_tokens_from_iterator	mindspore.dataset.text.Lookup	diff
torchtext.data.functional.sentencepiece_numericalizer	mindspore.dataset.text.SentencePieceTokenizer	diff
torchtext.data.functional.sentencepiece_tokenizer	mindspore.dataset.text.SentencePieceTokenizer	diff
torchtext.data.functional.simple_space_split	mindspore.dataset.text.WhitespaceTokenizer	diff
torchtext.data.utils.ngrams_iterator	mindspore.dataset.text.Ngram	diff
torchtext.datasets.AG_NEWS	mindspore.dataset.AGNewsDataset	diff
torchtext.datasets.AmazonReviewFull	mindspore.dataset.AmazonReviewDataset	diff
torchtext.datasets.AmazonReviewPolarity	mindspore.dataset.AmazonReviewDataset	diff
torchtext.datasets.CoNLL2000Chunking	mindspore.dataset.CoNLL2000Dataset	diff
torchtext.datasets.DBpedia	mindspore.dataset.DBpediaDataset	diff
torchtext.datasets.IMDB	mindspore.dataset.IMDBDataset	diff
torchtext.datasets.IWSLT2016	mindspore.dataset.IWSLT2016Dataset	diff
torchtext.datasets.IWSLT2017	mindspore.dataset.IWSLT2017Dataset	diff
torchtext.datasets.PennTreebank	mindspore.dataset.PennTreebankDataset	diff
torchtext.datasets.SogouNews	mindspore.dataset.SogouNewsDataset	diff
torchtext.datasets.SQuAD1	mindspore.dataset.SQuADDataset	diff
torchtext.datasets.SQuAD2	mindspore.dataset.SQuADDataset	diff
torchtext.datasets.UDPOS	mindspore.dataset.UDPOSDataset	diff
torchtext.datasets.WikiText103	mindspore.dataset.WikiTextDataset	diff
torchtext.datasets.WikiText2	mindspore.dataset.WikiTextDataset	diff
torchtext.datasets.YahooAnswers	mindspore.dataset.YahooAnswersDataset	diff
torchtext.datasets.YelpReviewFull	mindspore.dataset.YelpReviewDataset	diff
torchtext.datasets.YelpReviewPolarity	mindspore.dataset.YelpReviewDataset	diff

torchvision

The current API mapping table corresponds to PyTorch version 1.8.1, and Python version is no higher than Python 3.9.

TorchVision 0.9.1 APIs	MindSpore APIs	Descriptions
torchvision.datasets.CelebA	mindspore.dataset.CelebADataset	diff
torchvision.datasets.Cityscapes	mindspore.dataset.CityscapesDataset	diff
torchvision.datasets.CIFAR10	mindspore.dataset.Cifar10Dataset	diff
torchvision.datasets.CIFAR100	mindspore.dataset.Cifar100Dataset	diff
torchvision.datasets.CocoDetection	mindspore.dataset.CocoDataset	diff
torchvision.datasets.ImageFolder	mindspore.dataset.ImageFolderDataset	diff
torchvision.datasets.MNIST	mindspore.dataset.MnistDataset	diff
torchvision.datasets.VOCDetection	mindspore.dataset.VOCDataset	diff
torchvision.datasets.VOCSegmentation	mindspore.dataset.VOCDataset	diff
torchvision.ops.nms	mindspore.ops.NMSWithMask	diff
torchvision.ops.roi_align	mindspore.ops.ROIAlign	diff
torchvision.transforms.CenterCrop	mindspore.dataset.vision.CenterCrop	Consistent
torchvision.transforms.ColorJitter	mindspore.dataset.vision.RandomColorAdjust	Consistent
torchvision.transforms.Compose	mindspore.dataset.transforms.Compose	Consistent
torchvision.transforms.ConvertImageDtype	mindspore.dataset.transforms.TypeCast	diff
torchvision.transforms.FiveCrop	mindspore.dataset.vision.FiveCrop	Consistent
torchvision.transforms.GaussianBlur	mindspore.dataset.vision.GaussianBlur	Consistent
torchvision.transforms.Grayscale	mindspore.dataset.vision.Grayscale	Consistent
torchvision.transforms.LinearTransformation	mindspore.dataset.vision.LinearTransformation	Consistent
torchvision.transforms.Normalize	mindspore.dataset.vision.Normalize	diff
torchvision.transforms.Pad	mindspore.dataset.vision.Pad	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.RandomAffine	mindspore.dataset.vision.RandomAffine	diff
torchvision.transforms.RandomApply	mindspore.dataset.transforms.RandomApply	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.RandomChoice	mindspore.dataset.transforms.RandomChoice	Consistent
torchvision.transforms.RandomCrop	mindspore.dataset.vision.RandomCrop	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.RandomGrayscale	mindspore.dataset.vision.RandomGrayscale	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.RandomHorizontalFlip	mindspore.dataset.vision.RandomHorizontalFlip	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.RandomOrder	mindspore.dataset.transforms.RandomOrder	Consistent
torchvision.transforms.RandomPerspective	mindspore.dataset.vision.RandomPerspective	diff
torchvision.transforms.RandomResizedCrop	mindspore.dataset.vision.RandomResizedCrop	diff
torchvision.transforms.RandomRotation	mindspore.dataset.vision.RandomRotation	diff
torchvision.transforms.RandomVerticalFlip	mindspore.dataset.vision.RandomVerticalFlip	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.Resize	mindspore.dataset.vision.Resize	Consistent
torchvision.transforms.TenCrop	mindspore.dataset.vision.TenCrop	The functions are consistent, but the parameter names are inconsistent.
torchvision.transforms.ToPILImage	mindspore.dataset.vision.ToPIL	diff
torchvision.transforms.ToTensor	mindspore.dataset.vision.ToTensor	diff
torchvision.ops.deform_conv2d	mindspore.ops.deformable_conv2d	diff