Operation Reference¶
Operations
- A
- Abs
- Acos
- Acosh
- ActivityRegularization
- Adadelta
- Adafactor
- Adagrad
- Adam
- Adamax
- Adamw
- AdaptiveAveragePooling1D
- AdaptiveAveragePooling2D
- AdaptiveAveragePooling3D
- AdaptiveGradClipState
- AdaptiveMaxPooling1D
- AdaptiveMaxPooling2D
- AdaptiveMaxPooling3D
- Add
- AddDecayedWeightsState
- AddNoiseState
- AdditiveAttention
- All
- AllClose
- Alphadropout
- Angle
- Any
- Append
- ApplyAlongAxis
- ApplyIfFiniteState
- Arange
- ArgMax
- ArgMin
- ArgPartition
- ArgSort
- ArgWhere
- Array
- ArrayEqual
- ArrayEquiv
- ArrayLike
- ArraySplit
- AsArray
- Asin
- Asinh
- AssertClose
- Astype
- Atan
- Atan2
- Atanh
- Atleast1d
- Atleast2d
- Atleast3d
- Atom
- Attention
- AugMix
- AutoContrast
- AuxOutput
- Average
- AveragePooling1D
- AveragePooling2D
- AveragePooling3D
- AvgPool1d
- AvgPool2d
- AvgPool3d
- Avgpool
- B
- BLOCK_M
- Bartlett
- BatchNorm
- BatchNormalization
- Bernoulli
- Bfloat16
- Bidirectional
- Bilinear
- Bincount
- BitwiseAnd
- BitwiseCount
- BitwiseInvert
- BitwiseLeftShift
- BitwiseNot
- BitwiseOr
- BitwiseRightShift
- BitwiseXor
- Blackman
- Block
- Bool
- BroadcastArrays
- BroadcastShapes
- BroadcastTo
- C
- CTCLoss
- Cache
- Call
- CanCast
- Cast
- CastBool
- CastByte
- CastChar
- CastDouble
- CastFloat
- CastHalf
- CastInt
- CastLong
- CastShort
- CategoryEncoding
- Cbrt
- Ceil
- Celu
- CenterCrop
- Choose
- Clamp
- Clip
- ClipByGlobalNormState
- ClipGradNorm
- ClipState
- ColumnStack
- Compile
- Complex128
- Complex64
- ComplexWarning
- Compress
- ComputeCv
- Concat
- Concatenate
- ConditionallyMaskState
- ConditionallyTransformState
- Conj
- Conjugate
- Constant
- ControlVariate
- Conv
- Conv1DTranspose
- Conv1d
- Conv2DTranspose
- Conv2d
- Conv3DTranspose
- Conv3d
- ConvLSTM1D
- ConvLSTM2D
- ConvLSTM3D
- ConvT
- ConvTranspose1d
- ConvTranspose2d
- ConvTranspose3d
- Convolution1D
- Convolution1DTranspose
- Convolution2D
- Convolution2DTranspose
- Convolution3D
- Convolution3DTranspose
- Convolve
- Copy
- Copysign
- Corrcoef
- Correlate
- Cos
- Cosh
- CosineAnnealingLR
- CosineSimilarity
- CountNonzero
- Cov
- Cropping1D
- Cropping2D
- Cropping3D
- Cross
- CrossEntropyLoss
- Csingle
- CudaAvailable
- CumSum
- Cumprod
- CumulativeProd
- CumulativeSum
- CutMix
- CvExpectedValue
- CvState
- DataLoader
- DataParallel
- Deg2rad
- Degrees
- Delete
- Dense
- DepthwiseConv1D
- DepthwiseConv2D
- Det
- Device
- Diag
- DiagIndices
- DiagIndicesFrom
- Diagflat
- Diagonal
- Diff
- Digitize
- Discretization
- Div
- Divide
- Divmod
- Dot
- Double
- Dropout
- Dropout2d
- Dropout3d
- Dsplit
- Dstack
- Dtype
- E
- Ediff1d
- Einsum
- EinsumDense
- EinsumPath
- Elu
- EmaState
- Embedding
- Empty
- EmptyLike
- Equal
- Equalization
- Erf
- EulerGamma
- Exp
- Exp2
- ExpandDims
- Expm1
- Extract
- Eye
- Fabs
- FillDiagonal
- Finfo
- Fix
- Flatnonsense
- Flatnonzero
- Flatten
- FlaxLayer
- Flexible
- Flip
- Fliplr
- Flipud
- Float16
- Float32
- Float4_e2m1fn
- Float64
- Float8_e3m4
- Float8_e4m3
- Float8_e4m3b11fnuz
- Float8_e4m3fn
- Float8_e4m3fnuz
- Float8_e5m2
- Float8_e5m2fnuz
- Float8_e8m0fnu
- FloatPower
- Float
- Floating
- Floor
- FloorDivide
- Fmax
- Fmin
- Fmod
- Frexp
- FromBuffer
- FromDlpack
- FromFile
- FromFunction
- FromIter
- FromPyFunc
- FromString
- Full
- FullLike
- FusedMultiplyAdd
- GELU
- GRU
- Gather
- GaussianDropout
- GaussianNoise
- Gaussiannll
- Gcd
- Generic
- Geomspace
- GetPrintoptions
- GlobalAveragePooling1D
- GlobalAveragePooling2D
- GlobalAveragePooling3D
- GlobalAvgPool1D
- GlobalAvgPool2D
- GlobalAvgPool3D
- GlobalMaxPool1D
- GlobalMaxPool2D
- GlobalMaxPool3D
- GlobalMaxPooling1D
- GlobalMaxPooling2D
- GlobalMaxPooling3D
- Glu
- Gradient
- Grayscale
- Greater
- GreaterEqual
- Group
- GroupNormalization
- GroupQueryAttention
- Groupnorm
- Grucell
- Hamming
- Hanning
- HardSigmoid
- HardSwish
- Hardshrink
- Hardtanh
- HashedCrossing
- Hashing
- Heaviside
- Histogram
- Histogram2d
- HistogramBinEdges
- Histogramdd
- Hsplit
- Hstack
- Hypot
- I0
- INTEGER
- Identity
- Imag
- Indices
- InjectHyperparamsState
- InjectStatefulHyperparamsState
- Inner
- InputLayer
- InputSpec
- Insert
- Instancenorm
- Int16
- Int32
- Int64
- Int8
- IntegerLookup
- Interp
- Intersect1d
- Invert
- IsComplex
- IsComplexObj
- IsDtype
- IsFinite
- IsIn
- IsInf
- IsNaN
- IsNegInf
- IsPosInf
- IsReal
- IsRealObj
- IsScalar
- IsSubDtype
- Isclose
- Iterable
- Ix
- JaxLayer
- Kaiser
- Kldiv
- Kron
- L1
- LOWER_RIGHT
- LSTM
- Lambda
- Layer
- LayerNorm
- LayerNormalization
- LazyConvTranspose1d
- LazyConvTranspose2d
- LazyConvTranspose3d
- Lcm
- Ldexp
- LeakyRelu
- LeftShift
- Less
- LessEqual
- Lexsort
- Linear
- LinearT
- Linspace
- Lion
- List
- Load
- Log
- Log10
- Log1p
- Log2
- LogSoftmax
- Logaddexp
- Logaddexp2
- Logcosh
- LogicalAnd
- LogicalNot
- LogicalOr
- LogicalXor
- Logsigmoid
- Logspace
- Logsumexp
- Lstmcell
- M
- MA
- MSELoss
- MarginRankingLoss
- Marginranking
- MaskIndices
- MaskedNode
- MaskedState
- Masking
- MatMul
- MatrixTranspose
- Matvec
- Max
- MaxNumBoundingBoxes
- MaxPool1d
- MaxPool2d
- MaxPool3d
- MaxPooling1D
- MaxPooling2D
- MaxPooling3D
- MaxUnpool1d
- MaxUnpool2d
- MaxUnpool3d
- Maximum
- Maxpool
- Mean
- Median
- MelSpectrogram
- Meshgrid
- Mgrid
- Min
- Minimum
- Mish
- MixUp
- Mod
- Modf
- Module
- ModuleList
- Moveaxis
- Mse
- Mul
- MultiSteps
- MultiStepsState
- MultiTransformState
- MultiheadAttention
- Multinomial
- Multiply
- NONE
- Names
- Nan
- NanArgMax
- NanArgMin
- NanCumprod
- NanCumsum
- NanMax
- NanMean
- NanMedian
- NanMin
- NanPercentile
- NanProd
- NanToNum
- Nll
- NonNegativeParamsState
- NonZero
- Nop
- Normalization
- Normalize
- Object
- OneHot
- Ones
- OptState
- OptaxTest
- Optimizer
- Optional
- PRelu
- Pad
- Param
- Parameter
- ParameterList
- PartitionState
- Permute
- Pipeline
- Poly
- Polyint
- Polymul
- Polysub
- Polyval
- Positive
- Pow
- Power
- Printoptions
- Prod
- PromoteTypes
- Ptp
- Put
- PutAlongAxis
- Quantile
- RMSNormalization
- RMSprop
- RNN
- RNNBase
- RPC_AVAILABLE
- R
- Rad2deg
- Radians
- RandAugment
- RandInt
- RandomBrightness
- RandomColorDegeneration
- RandomColorJitter
- RandomContrast
- RandomCrop
- RandomElasticTransform
- RandomErasing
- RandomFlip
- RandomGaussianBlur
- RandomGrayscale
- RandomHorizontalFlip
- RandomHue
- RandomInvert
- RandomPerspective
- RandomPosterization
- RandomRotation
- RandomSaturation
- RandomSharpness
- RandomShear
- RandomTranslation
- RandomVerticalFlip
- RandomZoom
- Ravel
- RavelMultiIndex
- ReLU
- Real
- Reciprocal
- Relu6
- Remainder
- Repeat
- RepeatVector
- Rescaling
- Reshape
- Resize
- Resizing
- ResultType
- ReversibleEmbedding
- RightShift
- Rint
- Rmsnorm
- Rnncellbase
- Roll
- Rollaxis
- Roots
- Rot90
- Round
- SGD
- STFT
- STFTSpectrogram
- STRING
- Save
- Savez
- ScaleByTrustRatioState
- ScaleState
- Scan
- Scatter
- SearchSorted
- Select
- Selu
- SeparableConv1D
- SeparableConv2D
- SeparableConvolution1D
- SeparableConvolution2D
- Sequential
- SetDiff1d
- SetPrintoptions
- SetXor1d
- Shape
- ShouldSkipUpdateFunction
- SiLU
- Sigmoid
- Sign
- Signbit
- SignedInteger
- SimpleRNN
- SimpleRNNCell
- Sin
- Sinc
- Single
- Sinh
- Size
- Slice
- Smoothl1
- SnapshotState
- Softmax
- Softmax2d
- Softmin
- Softplus
- Softshrink
- Softsign
- Solarization
- Sort
- SortComplex
- Spacing
- Sparsemax
- SpatialDropout1D
- SpatialDropout2D
- SpatialDropout3D
- SpectralNorm
- SpectralNormalization
- Split
- Sqrt
- Square
- Squeeze
- Stack
- StackedRNNCells
- Std
- StepLR
- StringLookup
- Sub
- Subtract
- Sum
- Swapaxes
- Synchronize
- T
- TFSMLayer
- TUPLE
- T_module
- Take
- TakeAlongAxis
- Tan
- Tanh
- Tanhshrink
- Tensor
- TensorType
- Tensordot
- TextVectorization
- Threshold
- Tile
- TimeDistributed
- ToTensor
- TopK
- TorchFunctional
- TorchLoad
- TorchModuleWrapper
- TorchSave
- Trace
- TraceState
- Transformer
- TransformerDecoderLayer
- TransformerEncoderLayer
- Transformerdecoder
- Transformerencoder
- Transpose
- Trapezoid
- Tri
- Tril
- TrilIndices
- TrilIndicesFrom
- TrimZeros
- Triu
- TriuIndices
- TriuIndicesFrom
- TrueDivide
- Trunc
- UInt8
- Ufunc
- Uint
- Uint16
- Uint2
- Uint32
- Uint4
- Uint64
- Unflatten
- Unfold
- Union1d
- Unique
- UniqueAll
- UniqueCounts
- UniqueInverse
- UniqueValues
- UnitNormalization
- UnpackBits
- UnravelIndex
- UnsignedInteger
- Unsqueeze
- Unstack
- Unwrap
- UpSampling1D
- UpSampling2D
- UpSampling3D
- UpdateCvState
- Upsample
- V
- Vander
- Var
- Variable
- Vdot
- Vecdot
- Vecmat
- Vectorize
- View
- Vsplit
- Vstack
- WeightNorm
- Where
- WrappedSchedule
- Wrapper
- ZeroNansState
- ZeroPadding1D
- ZeroPadding2D
- ZeroPadding3D
- Zeros
- ZerosLike
- __array_namespace_info__
- absolute
- activation
- activation_relu_or_gelu
- adaptive_average_pool
- adaptive_grad_clip
- adaptive_max_pool
- add_decayed_weights
- add_noise
- add_scale
- addressable_data
- align_corners
- alpha
- amax
- amin
- apply
- apply_along_axis
- apply_gradients
- apply_if_finite
- apply_mask
- apply_over_axes
- approximate
- arccos
- arccosh
- arcsin
- arcsinh
- arctan
- arctan2
- arctanh
- args
- array_equal
- array_equiv
- array_split
- as_string
- assert_equal
- associative_scan
- at
- atleast_1d
- atleast_2d
- atleast_3d
- attr
- average_pool
- axis
- base
- batch_normalization
- batch_shape
- beta
- bias
- bias_correction
- bias_hh
- bias_ih
- bias_k
- bias_v
- binary_cross_entropy
- binomial
- bits
- bitwise_and
- bitwise_invert
- bitwise_left_shift
- bitwise_not
- bitwise_or
- bitwise_right_shift
- bitwise_xor
- bool
- broadcast
- broadcast_arrays
- broadcast_shapes
- broadcast_to
- byte
- cached
- can_cast
- cast_like
- categorical
- categorical_crossentropy
- cdouble
- ceil_mode
- chain
- char
- character
- children
- chisquare
- choice
- cholesky
- cholesky_inverse
- clear
- clip_by_block_rms
- clip_by_global_norm
- clip_grad_norm
- close
- column_stack
- complex
- complexfloating
- compute_mask
- cond
- conditionally_mask
- conditionally_transform
- constant
- constant_schedule
- control_delta_method
- control_variates_jacobians
- conv1d_input
- conv1d_weight
- conv2d_input
- conv2d_weight
- conv3d_input
- conv3d_weight
- conv_transpose
- conv_transpose1d
- conv_transpose2d
- conv_transpose3d
- convert_to_numpy
- convert_to_tensor
- convex_kl_divergence
- copy_to_host_async
- cosine_decay_schedule
- cosine_distance
- cosine_onecycle_schedule
- cosine_similarity
- count
- count_nonzero
- cpu
- create_mask
- cross_entropy
- ctc_decode
- ctc_loss
- ctc_loss_with_forward_probs
- cumulative_prod
- cumulative_sum
- custom_gradient
- d_model
- data
- data_ptr
- decoder
- depthwise_conv
- deserialize
- device_ids
- device_mesh
- device_type
- diag_indices
- diag_indices_from
- digamma
- dilation
- dim
- dims
- dirichlet
- div_value
- divide_no_nan
- dot_product_attention
- downscale_factor
- dropout1
- dropout2
- dropout3
- eig
- eigh
- eigvals
- eigvalsh
- einsum_path
- ema
- embedding_dim
- empty_like
- enable_grad
- encoder
- end_dim
- eps
- erfc
- erfcx
- erfinv
- estimate_control_variate_coefficients
- euler_gamma
- eval
- expand_dims
- expit
- exponential
- exponential_decay
- extend
- extract_sequences
- eye
- f
- fft
- fft2
- fftfreq
- fftn
- fftshift
- fill_diagonal
- filters
- find_duplicates
- flatten_parameters
- float
- floor_divide
- fn
- fori_loop
- forward
- freeze
- from_dlpack
- full_like
- gamma
- gammainc
- gammaincc
- gammaln
- gaussian_nll_loss
- geometric
- get
- get_all_with_path
- get_item
- get_parameter
- get_printoptions
- glorot_normal
- glorot_uniform
- grad
- greater_equal
- group_size
- groups
- gumbel
- half
- hard_shrink
- hard_sigmoid
- hard_silu
- hard_swish
- hard_tanh
- he_normal
- he_uniform
- head
- head_dim
- hfft
- hidden_size
- hinge_loss
- histogram_bin_edges
- huber_loss
- ifft
- ifft2
- ifftn
- ifftshift
- ihfft
- iinfo
- in1_features
- in2_features
- in_features
- in_top_k
- inexact
- inf
- init
- inject_hyperparams
- inject_stateful_hyperparams
- inplace
- int
- int2
- int4
- int
- interpolate
- inv
- invert_permutation
- irfft
- irfft2
- irfftn
- is_tensor
- istft
- item
- items
- itemsize
- jit
- join
- join_schedules
- jvp
- k
- kaiming_normal
- kaiming_uniform
- keep_params_nonnegative
- kernel_size
- key
- keys
- kl_divergence
- l1_loss
- l2_loss
- label_smoothing
- lambd
- laplace
- layer_norm
- layers
- leaky_relu
- left_shift
- less_equal
- linear1
- linear2
- linear_onecycle_schedule
- linear_schedule
- linear_to_mel_weight_matrix
- load_state_dict
- log_cosh
- log_ndtr
- log_prob
- log_sigmoid
- log_softmax
- logdet
- logger
- logical_and
- logical_not
- logical_or
- logical_xor
- logistic
- logit
- lognormal
- long
- lower
- lstsq
- lu
- lu_factor
- map
- map_params
- margin
- margin_ranking_loss
- mask
- mask_check
- mask_indices
- mask_mod
- masked
- matrix_norm
- matrix_power
- matrix_rank
- matrix_transpose
- max_norm
- max_pool
- max_pool1d
- max_pool2d
- max_pool3d
- max_val
- measure_valued_estimation_mean
- measure_valued_estimation_std
- measure_valued_jacobians
- merge_masks
- metadata
- min_val
- mode
- modules
- moments
- momentum
- moving_avg_baseline
- mse_loss
- multi_dot
- multi_transform
- multivariate_normal
- n
- name
- named_chain
- named_modules
- named_parameters
- nan_to_num
- nbytes
- ndarray
- ndim
- negative
- negative_slope
- nextafter
- nll_loss
- no_grad
- nonlinearity
- norm
- norm1
- norm2
- norm3
- norm_first
- norm_type
- normal
- not_equal
- ntxent
- num_embeddings
- num_features
- num_groups
- num_heads
- num_layers
- number
- object
- one_hot
- ones
- ones_like
- orthogonal
- orthogonal
- out_features
- out_proj
- outer
- output_device
- output_ratio
- output_size
- p
- padding
- padding_idx
- parameters
- parameters_to_vector
- params
- pareto
- partition
- pathwise_jacobians
- per_example_global_norm_clip
- per_example_layer_norm_clip
- permutation
- permute_dims
- pi
- piecewise
- piecewise_constant_schedule
- piecewise_interpolate_schedule
- pinv
- place
- poisson
- polar
- polydiv
- polynomial_schedule
- pop
- popitem
- predict
- promote_types
- psi
- psnr
- qr
- quantize
- randn
- random_like
- rank
- ravel_multi_index
- rayleigh
- register_buffer
- register_parameter
- remove
- remove_axis
- replace
- reset
- reshape_weight_to_matrix
- result_type
- rfft
- rfft2
- rfftfreq
- rfftn
- right_shift
- rms_norm
- rms_normalization
- rsqrt
- safe_softmax_cross_entropy
- saturate_cast
- savez_compressed
- scale
- scale_factor
- scale_grad_by_freq
- scaled_dot_product_attention
- scatter_update
- score_function_jacobians
- seed
- segment_max
- segment_sum
- selective_transform
- separable_conv
- serialize
- set
- set_printoptions
- sgdr_schedule
- short
- shuffle
- sigmoid_binary_cross_entropy
- sigmoid_focal_loss
- skip_large_updates
- skip_not_finite
- slice_update
- slogdet
- smooth_l1_loss
- smooth_labels
- snapshot
- soft_shrink
- softmax_cross_entropy
- softmax_cross_entropy_with_integer_labels
- solve
- solve_triangular
- sparse
- sparse
- sparse_categorical_crossentropy
- sparse_plus
- sparse_sigmoid
- split_key_like
- squared_error
- squareplus
- start_dim
- state
- state_dict
- static_graph
- step
- stop_gradient
- stride
- strides
- svd
- svdvals
- swap
- switch
- take_along_axis
- tanh_shrink
- tensorinv
- tensorsolve
- test_import
- threshold
- to_dense
- to_device
- tolist
- top_k
- train
- training
- tree_add_scalar_mul
- tree_l1_norm
- tree_l2_norm
- tree_linf_norm
- tree_map_params
- tree_scalar_mul
- triangular
- tril_indices
- tril_indices_from
- trim_zeros
- triu_indices
- triu_indices_from
- true_divide
- trunc_normal
- truncated_normal
- type
- unflattened_size
- uniform
- unique_all
- unique_counts
- unique_inverse
- unique_values
- unravel_index
- unwrap_random_key_data
- update
- update_infinity_moment
- update_moment
- update_moment_per_elem_norm
- upper
- upsample_bilinear
- upsample_nearest
- upscale_factor
- value
- value_and_grad
- values
- variant
- vector_norm
- vectorized_map
- view_as_complex
- view_as_real
- vmap
- wald
- warmup_constant_schedule
- warmup_cosine_decay_schedule
- warmup_exponential_decay_schedule
- weight
- weight_hh
- weight_ih
- while_loop
- xavier_normal
- xavier_uniform
- xlog1py
- zero_grad
- zero_nans
- zeros
- zeros_like