Fix issue with Metric/Loss wrapper

Author: t-kalinowski

NEWS.md

@@ -13,6 +13,9 @@
 
 - `%*%` now dispatches to `op_matmul()` for tensorflow tensors, which
   has relaxed shape constraints compared to `tf$matmul()`.
+  
+- Fixed an issue where calling a `Metric` and `Loss` object 
+  with unnamed arguments would error.
 
 ## Added compatibility with Keras v3.8.0. User-facing changes:
 

R/losses.R

@@ -1962,7 +1962,8 @@ py_to_r_wrapper.keras.src.losses.loss.Loss <- function(x) {
   as.function.default(c(formals(x), quote({
     args <- capture_args(list(y_true = as_py_array,
                               y_pred = as_py_array,
-                              sample_weight = as_py_array))
+                              sample_weight = as_py_array),
+                         enforce_all_dots_named = FALSE)
     do.call(x, args)
   })))
 }

R/metrics.R

@@ -2978,6 +2978,13 @@ function (y_true, y_pred, from_logits = FALSE, ignore_class = NULL,
 #' ```
 #'
 #' ```{r}
+#' # calling a metric directly is equivalent to calling
+#' # m$update_state(); m$result()
+#' m <- metric_mean()
+#' m(c(1, 3, 5, 7))
+#' ```
+#'
+#' ```{r}
 #' m$reset_state()
 #' m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0))
 #' m$result()