Merge pull request #73 from mlverse/feature/evaluate

Implements `evaluate`

NAMESPACE

@@ -17,16 +17,22 @@ S3method(as_dataset,numeric)
 S3method(as_dataset,torch_tensor)
 S3method(as_iterator,device_dataloader)
 S3method(fit,luz_module_generator)
+S3method(get_metrics,luz_context)
+S3method(get_metrics,luz_module_evaluation)
+S3method(get_metrics,luz_module_fitted)
 S3method(plot,lr_records)
 S3method(plot,luz_module_fitted)
 S3method(predict,luz_module_fitted)
 S3method(print,lr_records)
+S3method(print,luz_module_evaluation)
 S3method(print,luz_module_fitted)
 S3method(print,luz_module_generator)
 export("%>%")
 export(accelerator)
 export(as_dataloader)
+export(evaluate)
 export(fit)
+export(get_metrics)
 export(lr_finder)
 export(luz_callback)
 export(luz_callback_csv_logger)

NEWS.md

@@ -10,6 +10,8 @@
 * We now handle different kinds of data arguments passed to `fit` using the `as_dataloader()` method (#66).
 * `valid_data` can now be scalar value indicating the proportion of `data` that will be used for fitting. This only works if `data` is a torch dataset or a list. (#69)
 * You can now supply `dataloader_options` to `fit` to pass additional information to `as_dataloader()`. (#71)
+* Refactored the `ctx` object to make it safer and avoid returing it in the output. (#73)
+* Implemented the `evaluate` function allowing users to get metrics from a model in a new datase. (#73)
 
 # luz 0.1.0
 

R/as_dataloader.R

@@ -91,7 +91,7 @@ as_dataset <- function(x, ...) {
 
 #' @export
 as_dataset.default <- function(x, ...) {
-  rlang::abort(sprinf("Can't convert object with class '%s' to a torch dataset.", class(x)[1]))
+  rlang::abort(sprintf("Can't convert object with class '%s' to a torch dataset.", class(x)[1]))
 }
 
 #' @export

R/callbacks-profile.R

@@ -95,9 +95,9 @@ luz_callback_profile <- luz_callback(
 )
 
 get_total_time <- function(x) {
-  unlist(x$ctx$records$profile$fit)
+  unlist(x$records$profile$fit)
 }
 
 get_average_time <- function(x, what) {
-  mean(unlist(x$ctx$records$profile[[what]]))
+  mean(unlist(x$records$profile[[what]]))
 }

R/callbacks.R

@@ -37,7 +37,17 @@ default_callbacks <- function() {
 
 default_predict_callbacks <- function() {
   list(
-    luz_callback_progress()
+    luz_callback_progress(),
+    luz_callback_interrupt()
+  )
+}
+
+default_evaluate_callbacks <- function() {
+  list(
+    luz_callback_profile(),
+    luz_callback_metrics(),
+    luz_callback_progress(),
+    luz_callback_interrupt()
   )
 }
 
@@ -535,7 +545,7 @@ luz_callback_lr_scheduler <- luz_callback(
     if (is.null(self$opt_name) && (length(ctx$optimizers) == 1))
       self$opt_name <- names(ctx$optimizers)
     else
-      rlang::abort("An optimizer name was not supported and your model has multiple optimizers")
+      rlang::abort("An optimizer name was not supplied and your model has multiple optimizers")
 
     if (!self$opt_name %in% names(ctx$optimizers))
       rlang::abort(glue::glue("opt_name '{self$opt_name}' not found in ctx$optimizers."))