TorchScript: statically-typed graph representation of PyTorch models

Traces an op sequence into a static graph (shapes frozen at each node), then verifies by replay.

Usage

morie_geron_torchscript(model, example_inputs)

Arguments

model: List of ops (see morie_geron_trace).
example_inputs: Representative input.

Value

list with graph, output, replay, max_diff, n_nodes, shapes, estimate, n, method.

Examples

W <- diag(2)
morie_geron_torchscript(list(list(kind = "linear", param = W),
    list(kind = "relu")), matrix(c(1, -1), 1, 2))
#> $graph
#> $graph[[1]]
#> $graph[[1]]$index
#> [1] 0
#> 
#> $graph[[1]]$kind
#> [1] "linear"
#> 
#> $graph[[1]]$in_shape
#> [1] 1 2
#> 
#> $graph[[1]]$out_shape
#> [1] 1 2
#> 
#> $graph[[1]]$op
#> $graph[[1]]$op$kind
#> [1] "linear"
#> 
#> $graph[[1]]$op$param
#>      [,1] [,2]
#> [1,]    1    0
#> [2,]    0    1
#> 
#> 
#> 
#> $graph[[2]]
#> $graph[[2]]$index
#> [1] 1
#> 
#> $graph[[2]]$kind
#> [1] "relu"
#> 
#> $graph[[2]]$in_shape
#> [1] 1 2
#> 
#> $graph[[2]]$out_shape
#> [1] 1 2
#> 
#> $graph[[2]]$op
#> $graph[[2]]$op$kind
#> [1] "relu"
#> 
#> 
#> 
#> 
#> $output
#>      [,1] [,2]
#> [1,]    1    0
#> 
#> $replay
#>      [,1] [,2]
#> [1,]    1    0
#> 
#> $max_diff
#> [1] 0
#> 
#> $n_nodes
#> [1] 2
#> 
#> $shapes
#> $shapes[[1]]
#> $shapes[[1]][[1]]
#> [1] 1 2
#> 
#> $shapes[[1]][[2]]
#> [1] 1 2
#> 
#> 
#> $shapes[[2]]
#> $shapes[[2]][[1]]
#> [1] 1 2
#> 
#> $shapes[[2]][[2]]
#> [1] 1 2
#> 
#> 
#> 
#> $estimate
#> [1] 0
#> 
#> $n
#> [1] 1
#> 
#> $method
#> [1] "Shape-specialised tracing of an op sequence into a static graph, verified by replay"
#>