Ingress for a single lazy_tensor column.
Parameters
shape::integer()
The shape of the tensor, where the first dimension (batch) must beNA. When it is not specified, the lazy tensor input column needs to have a known shape.
Input and Output Channels
One input channel called "input" and one output channel called "output".
For an explanation see PipeOpTorch.
See also
Other PipeOps:
mlr_pipeops_nn_adaptive_avg_pool1d,
mlr_pipeops_nn_adaptive_avg_pool2d,
mlr_pipeops_nn_adaptive_avg_pool3d,
mlr_pipeops_nn_avg_pool1d,
mlr_pipeops_nn_avg_pool2d,
mlr_pipeops_nn_avg_pool3d,
mlr_pipeops_nn_batch_norm1d,
mlr_pipeops_nn_batch_norm2d,
mlr_pipeops_nn_batch_norm3d,
mlr_pipeops_nn_block,
mlr_pipeops_nn_celu,
mlr_pipeops_nn_conv1d,
mlr_pipeops_nn_conv2d,
mlr_pipeops_nn_conv3d,
mlr_pipeops_nn_conv_transpose1d,
mlr_pipeops_nn_conv_transpose2d,
mlr_pipeops_nn_conv_transpose3d,
mlr_pipeops_nn_dropout,
mlr_pipeops_nn_elu,
mlr_pipeops_nn_flatten,
mlr_pipeops_nn_gelu,
mlr_pipeops_nn_glu,
mlr_pipeops_nn_hardshrink,
mlr_pipeops_nn_hardsigmoid,
mlr_pipeops_nn_hardtanh,
mlr_pipeops_nn_head,
mlr_pipeops_nn_layer_norm,
mlr_pipeops_nn_leaky_relu,
mlr_pipeops_nn_linear,
mlr_pipeops_nn_log_sigmoid,
mlr_pipeops_nn_max_pool1d,
mlr_pipeops_nn_max_pool2d,
mlr_pipeops_nn_max_pool3d,
mlr_pipeops_nn_merge,
mlr_pipeops_nn_merge_cat,
mlr_pipeops_nn_merge_prod,
mlr_pipeops_nn_merge_sum,
mlr_pipeops_nn_prelu,
mlr_pipeops_nn_relu,
mlr_pipeops_nn_relu6,
mlr_pipeops_nn_reshape,
mlr_pipeops_nn_rrelu,
mlr_pipeops_nn_selu,
mlr_pipeops_nn_sigmoid,
mlr_pipeops_nn_softmax,
mlr_pipeops_nn_softplus,
mlr_pipeops_nn_softshrink,
mlr_pipeops_nn_softsign,
mlr_pipeops_nn_squeeze,
mlr_pipeops_nn_tanh,
mlr_pipeops_nn_tanhshrink,
mlr_pipeops_nn_threshold,
mlr_pipeops_nn_unsqueeze,
mlr_pipeops_torch_ingress,
mlr_pipeops_torch_ingress_categ,
mlr_pipeops_torch_ingress_num,
mlr_pipeops_torch_loss,
mlr_pipeops_torch_model,
mlr_pipeops_torch_model_classif,
mlr_pipeops_torch_model_regr
Other Graph Network:
ModelDescriptor(),
TorchIngressToken(),
mlr_learners_torch_model,
mlr_pipeops_module,
mlr_pipeops_torch,
mlr_pipeops_torch_ingress,
mlr_pipeops_torch_ingress_categ,
mlr_pipeops_torch_ingress_num,
model_descriptor_to_learner(),
model_descriptor_to_module(),
model_descriptor_union(),
nn_graph()
Super classes
mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressLazyTensor
Methods
Method new()
Creates a new instance of this R6 class.
Usage
PipeOpTorchIngressLazyTensor$new(
id = "torch_ingress_ltnsr",
param_vals = list()
)Arguments
id(
character(1))
Identifier of the resulting object.param_vals(
list())
List of hyperparameter settings, overwriting the hyperparameter settings that would otherwise be set during construction.
Examples
po_ingress = po("torch_ingress_ltnsr")
task = tsk("lazy_iris")
md = po_ingress$train(list(task))[[1L]]
ingress = md$ingress
x_batch = ingress[[1L]]$batchgetter(data = task$data(1, "x"), device = "cpu", cache = NULL)
x_batch
#> torch_tensor
#> 5.1000 3.5000 1.4000 0.2000
#> [ CPUFloatType{1,4} ]
# Now we try a lazy tensor with unknown shape, i.e. the shapes between the rows can differ
ds = dataset(
initialize = function() self$x = list(torch_randn(3, 10, 10), torch_randn(3, 8, 8)),
.getitem = function(i) list(x = self$x[[i]]),
.length = function() 2)()
task_unknown = as_task_regr(data.table(
x = as_lazy_tensor(ds, dataset_shapes = list(x = NULL)),
y = rnorm(2)
), target = "y", id = "example2")
# this task (as it is) can NOT be processed by PipeOpTorchIngressLazyTensor
# It therefore needs to be preprocessed
po_resize = po("trafo_resize", size = c(6, 6))
task_unknown_resize = po_resize$train(list(task_unknown))[[1L]]
# printing the transformed column still shows unknown shapes,
# because the preprocessing pipeop cannot infer them,
# however we know that the shape is now (3, 10, 10) for all rows
task_unknown_resize$data(1:2, "x")
#> x
#> <lazy_tensor>
#> 1: <tnsr[]>
#> 2: <tnsr[]>
po_ingress$param_set$set_values(shape = c(NA, 3, 6, 6))
md2 = po_ingress$train(list(task_unknown_resize))[[1L]]
ingress2 = md2$ingress
x_batch2 = ingress2[[1L]]$batchgetter(
data = task_unknown_resize$data(1:2, "x"),
device = "cpu",
cache = NULL
)
x_batch2
#> torch_tensor
#> (1,1,.,.) =
#> 0.1449 -0.7019 -0.3521 -0.0491 -1.7385 -0.5847
#> -0.8391 -1.4985 -0.6923 0.7996 0.4785 0.5215
#> 0.3497 -0.0709 1.2281 -0.4835 0.4081 0.0466
#> -0.2712 -1.6234 -0.3864 0.1636 1.1627 0.9997
#> -0.2890 0.1777 1.3893 -1.2745 1.0467 -0.7197
#> 0.0156 0.1764 0.6456 -0.0333 -0.4097 0.1578
#>
#> (2,1,.,.) =
#> 1.0986 -0.1841 0.8730 0.1146 0.9023 0.1112
#> 0.9094 0.3374 0.1289 0.2231 0.7119 -0.0117
#> 0.1584 0.8659 -0.1255 0.5785 -0.1748 1.2001
#> 0.0245 0.7896 0.2707 0.2722 1.1083 -0.7059
#> 0.1058 -0.3097 0.5502 0.6818 -0.7135 -0.4190
#> 1.0454 -0.5466 -0.3665 -0.2877 -0.1238 0.2257
#>
#> (1,2,.,.) =
#> -1.0097 0.8060 0.0781 -0.5706 0.6929 -1.1018
#> -0.6435 0.4388 0.3039 -0.9457 -0.7491 -0.7958
#> 0.2781 -0.2517 0.7829 -0.7085 -1.0625 0.7910
#> 0.4273 0.4935 0.2813 0.8131 -0.0913 -0.9181
#> 0.1782 0.1244 0.0531 0.6148 0.4526 -0.2287
#> -0.2312 1.2950 0.0757 -0.2358 -0.5711 1.0204
#>
#> (2,2,.,.) =
#> 1.0439e+00 -1.0122e-02 -6.1007e-01 -3.0304e-01 -9.2154e-02 -5.4351e-01
#> -4.2073e-01 3.3601e-02 -1.0326e-02 -5.7285e-01 -5.1729e-05 -8.3567e-01
#> -4.0730e-01 -1.1399e-01 -4.6574e-01 1.8683e-01 4.4185e-01 4.4938e-01
#> 1.8344e+00 -1.9484e-01 -8.9078e-01 7.7812e-01 6.4956e-01 -9.9950e-01
#> 1.0212e+00 -9.2776e-02 1.8537e-01 -7.1413e-01 -1.1227e-01 1.7768e-01
#> ... [the output was truncated (use n=-1 to disable)]
#> [ CPUFloatType{2,3,6,6} ]