Logs metrics and other model information in the tfevents file format. Assuming tensorboard is installed, result can be visualized with
Arguments
- logdir
A directory to where log will be written to.
- histograms
A boolean specifying if histograms of model weights should be logged. It can also be a character vector specifying the name of the parameters that should be logged (names are the same as
names(model$parameters)).- ...
Currently not used. For future expansion.
Examples
if (torch::torch_is_installed()) {
library(torch)
x <- torch_randn(1000, 10)
y <- torch_randn(1000, 1)
model <- nn_linear %>%
setup(loss = nnf_mse_loss, optimizer = optim_adam) %>%
set_hparams(in_features = 10, out_features = 1) %>%
set_opt_hparams(lr = 1e-4)
tmp <- tempfile()
model %>% fit(list(x, y), valid_data = 0.2, callbacks = list(
luz_callback_tfevents(tmp, histograms = TRUE)
))
}
#> A `luz_module_fitted`
#> ── Time ────────────────────────────────────────────────────────────────────────
#> • Total time: 1.6s
#> • Avg time per training epoch: 118ms
#>
#> ── Results ─────────────────────────────────────────────────────────────────────
#> Metrics observed in the last epoch.
#>
#> ℹ Training:
#> loss: 1.2572
#>
#> ── Model ───────────────────────────────────────────────────────────────────────
#> An `nn_module` containing 11 parameters.
#>
#> ── Parameters ──────────────────────────────────────────────────────────────────
#> • weight: Float [1:1, 1:10]
#> • bias: Float [1:1]