[R] Update the R for roxygen

Makefile

@@ -149,7 +149,7 @@ rcpplint:
 	python dmlc-core/scripts/lint.py mxnet-rcpp ${LINT_LANG} R-package/src
 
 rcppexport:
-	Rscript -e "require(mxnet); mxnet::mxnet.export(\"R-package/R\")"
+	Rscript -e "require(mxnet); mxnet::mxnet.export(\"R-package\")"
 
 roxygen:
 	Rscript -e "require(roxygen2); roxygen2::roxygenise(\"R-package\")"

R-package/NAMESPACE

@@ -1,3 +1,74 @@
-exportPattern("^[[:alpha:]]+")
+# Generated by roxygen2 (4.1.1): do not edit by hand
+
+S3method(Ops,MXNDArray)
+S3method(as.array,MXNDArray)
+S3method(as.matrix,MXNDArray)
+S3method(dim,MXNDArray)
+S3method(length,MXNDArray)
+S3method(predict,MXFeedForwardModel)
+S3method(print,MXNDArray)
+export(arguments)
+export(ctx)
+export(is.mx.context)
+export(is.mx.dataiter)
+export(is.mx.ndarray)
+export(is.mx.symbol)
+export(mx.apply)
+export(mx.callback.log.train.metric)
+export(mx.callback.save.checkpoint)
+export(mx.cpu)
+export(mx.ctx.default)
+export(mx.exec.backward)
+export(mx.exec.forward)
+export(mx.exec.update.arg.arrays)
+export(mx.exec.update.aux.arrays)
+export(mx.gpu)
+export(mx.init.create)
+export(mx.init.normal)
+export(mx.init.uniform)
+export(mx.io.ImageRecordIter)
+export(mx.io.MNISTIter)
+export(mx.io.arrayiter)
+export(mx.io.extract)
+export(mx.kv.create)
+export(mx.metric.accuracy)
+export(mx.model.FeedForward.create)
+export(mx.model.load)
+export(mx.model.save)
+export(mx.nd.array)
+export(mx.nd.clip)
+export(mx.nd.copyto)
+export(mx.nd.dot)
+export(mx.nd.load)
+export(mx.nd.ones)
+export(mx.nd.save)
+export(mx.nd.zeros)
+export(mx.opt.create)
+export(mx.opt.get.updater)
+export(mx.rnorm)
+export(mx.runif)
+export(mx.set.seed)
+export(mx.simple.bind)
+export(mx.symbol.Activation)
+export(mx.symbol.BatchNorm)
+export(mx.symbol.Concat)
+export(mx.symbol.Convolution)
+export(mx.symbol.Dropout)
+export(mx.symbol.ElementWiseSum)
+export(mx.symbol.Flatten)
+export(mx.symbol.FullyConnected)
+export(mx.symbol.Group)
+export(mx.symbol.LRN)
+export(mx.symbol.LeakyReLU)
+export(mx.symbol.Pooling)
+export(mx.symbol.Reshape)
+export(mx.symbol.SliceChannel)
+export(mx.symbol.Softmax)
+export(mx.symbol.Variable)
+export(mx.symbol.infer.shape)
+export(mx.symbol.load)
+export(mx.symbol.save)
+export(mxnet.export)
+export(outputs)
 import(Rcpp)
 import(methods)

R-package/R/callback.R

@@ -1,4 +1,4 @@
-# Log training metric each period
+#' Log training metric each period
 #' @export
 mx.callback.log.train.metric <- function(period) {
   function(iteration, nbatch, env) {
@@ -10,7 +10,10 @@ mx.callback.log.train.metric <- function(period) {
 }
 
 
-#
+#' Save checkpoint to files each period iteration.
+#'
+#' @param prefix The prefix of the model checkpoint.
+#'
 #' @export
 mx.callback.save.checkpoint <- function(prefix, period=1) {
   function(iteration, nbatch, env) {

R-package/R/context.R

@@ -7,6 +7,7 @@ init.context.default <- function() {
 #'
 #' @param new, optional takes \code{mx.cpu()} or \code{mx.gpu(id)}, new default ctx.
 #' @return The default context.
+#'
 #' @export
 mx.ctx.default <- function(new = NULL) {
   if (!is.null(new)) {
@@ -18,6 +19,7 @@ mx.ctx.default <- function(new = NULL) {
 #' Check if the type is mxnet context.
 #'
 #' @return Logical indicator
+#'
 #' @export
 is.mx.context <- function(x) {
   class(x) == "MXContext"
@@ -30,6 +32,7 @@ is.mx.context <- function(x) {
 #'     The device ID, this is meaningless for CPU, included for interface compatiblity.
 #' @return The CPU context.
 #' @name mx.cpu
+#'
 #' @export
 NULL
 
@@ -39,5 +42,6 @@ NULL
 #'     The GPU device ID, starts from 0.
 #' @return The GPU context.
 #' @name mx.gpu
+#'
 #' @export
 NULL

R-package/R/executor.R

@@ -1,5 +1,6 @@
-# Simple bind the symbol to executor,
-# with information from input shapes.
+#' Simple bind the symbol to executor,
+#' with information from input shapes.
+#'
 #' @export
 mx.simple.bind <- function(symbol, ctx, grad.req=FALSE, ...) {
   if (!is.MXSymbol(symbol)) stop("symbol need to be MXSymbol")
@@ -43,13 +44,15 @@ mx.exec.update.aux.arrays <- function(exec, arg.arrays, match.name=FALSE, skip.n
 
 #' Peform an forward on the executors
 #' This function will MUTATE the state of exec
+#'
 #' @export
 mx.exec.forward <- function(exec, is.train=TRUE) {
   exec$forward(is.train, list())
 }
 
 #' Peform an backward on the executors
 #' This function will MUTATE the state of exec
+#'
 #' @export
 mx.exec.backward <- function(exec, ...) {
   exec$backward(list(...))

R-package/R/initializer.R

@@ -1,6 +1,8 @@
 #' Internal default value initialization scheme.
+#'
 #' @param name the name of the variable.
 #' @param shape the shape of the array to be generated.
+#'
 mx.init.internal.default <- function(name, shape, ctx, allow.unknown=FALSE) {
   if (mx.util.str.endswith(name, "bias")) return (mx.nd.zeros(shape, ctx))
   if (mx.util.str.endswith(name, "gamma")) return (mx.nd.ones(shape, ctx))
@@ -27,7 +29,7 @@ mx.init.uniform <- function(scale) {
 
 #' Create a initializer that initialize the weight with normal(0, sd)
 #'
-#' @param scale The scale of uniform distribution
+#' @param sd The standard deviation of normal distribution
 #'
 #' @export
 mx.init.normal <- function(sd) {
@@ -39,7 +41,13 @@ mx.init.normal <- function(sd) {
   }
 }
 
-# Create initialization of argument  like arg.array
+#' Create initialization of argument  like arg.array
+#'
+#' @param initializer The initializer.
+#' @param shape.array named-list The shape of the weights
+#' @param ctx mx.context The context of the weights
+#' @param skip.unknown Whether skip the unknown weight types
+#' @export
 mx.init.create <- function(initializer, shape.array, ctx, skip.unknown=TRUE) {
   if (length(shape.array) == 0) return(list())
   names = names(shape.array)

R-package/R/io.R

@@ -3,10 +3,18 @@ is.MXDataIter <- function(x) {
   inherits(x, "Rcpp_MXArrayDataIter")
 }
 
+#' Judge if an object is mx.dataiter
+#'
+#' @return Logical indicator
+#'
+#' @export
+is.mx.dataiter <- is.MXDataIter
+
 #' Extract a certain field from DataIter.
+#'
 #' @export
 mx.io.extract <- function(iter, field) {
-  packer <- mx.nd.arraypacker.create()
+  packer <- mx.nd.arraypacker()
   iter$reset()
   while (iter$iter.next()) {
     dlist <- iter$value()
@@ -15,9 +23,11 @@ mx.io.extract <- function(iter, field) {
     oshape <- dim(data)
     packer$push(mx.nd.slice(data, 0, oshape[[1]] - padded))
   }
+  iter$reset()
   return(packer$get())
 }
 
+#
 #' Create MXDataIter compatible iterator from R's array
 #'
 #' @param data The data array.
@@ -26,11 +36,11 @@ mx.io.extract <- function(iter, field) {
 #' @param shuffle Whether shuffle the data
 #'
 #' @export
-mx.io.ArrayIter <- function(data, label=NULL,
+mx.io.arrayiter <- function(data, label=NULL,
                             batch.size=128,
                             shuffle=FALSE) {
-  mx.io.internal.ArrayIter.create(as.array(data),
-                                  as.array(label),
-                                  batch.size,
-                                  shuffle)
+  mx.io.internal.arrayiter(as.array(data),
+                           as.array(label),
+                           batch.size,
+                           shuffle)
 }

R-package/R/kvstore.R

@@ -1,3 +1,12 @@
 is.MXKVStore <- function(x) {
   inherits(x, "Rcpp_MXKVStore")
 }
+
+#' Create a mxnet KVStore.
+#'
+#' @param type string(default="local") The type of kvstore.
+#' @return The kvstore.
+#'
+#' @name mx.kv.create
+#' @export
+NULL

R-package/R/metric.R

@@ -16,7 +16,8 @@ mx.metric.custom <-function(name, feval) {
   return(ret)
 }
 
-# accuracy metric
+#' Accuracy metric
+#'
 #' @export
 mx.metric.accuracy <- mx.metric.custom("accuracy", function(label, pred) {
   ypred = max.col(as.array(pred), tie="first")

R-package/R/model.R

@@ -257,7 +257,7 @@ mx.model.init.iter <- function(X, y, batch.size, is.train) {
     shape <- dim(X)
     y <- c(1:shape[[1]]) * 0
   }
-  return(mx.io.ArrayIter(X, y, batch.size=batch.size, shuffle=is.train))
+  return(mx.io.arrayiter(X, y, batch.size=batch.size, shuffle=is.train))
 }
 
 #' Create a MXNet Feedforward neural net model with the specified training.
@@ -288,6 +288,7 @@ mx.model.init.iter <- function(X, y, batch.size, is.train) {
 #' @param kvstore string (default="local")
 #'     The parameter synchronization scheme in multiple devices.
 #' @return model A trained mxnet model.
+#'
 #' @export
 mx.model.FeedForward.create <-
 function(symbol, X, y=NULL, ctx=NULL,
@@ -333,6 +334,7 @@ function(symbol, X, y=NULL, ctx=NULL,
 #' @param X The dataset to predict.
 #' @param ctx mx.cpu() or mx.gpu(i) The device used to generate the prediction.
 #' @param array.batch.size The batch size used in batching. Only used when X is R's array.
+#'
 #' @export
 predict.MXFeedForwardModel <- function(model, X, ctx=NULL, array.batch.size=128) {
   if (is.null(ctx)) ctx <- mx.ctx.default()
@@ -343,7 +345,7 @@ predict.MXFeedForwardModel <- function(model, X, ctx=NULL, array.batch.size=128)
   pexec <- mx.simple.bind(model$symbol, ctx=ctx, data=dim(dlist$data), grad.req=FALSE)
   mx.exec.update.arg.arrays(pexec, model$arg.params, match.name=TRUE)
   mx.exec.update.aux.arrays(pexec, model$aux.params, match.name=TRUE)
-  packer <- mx.nd.arraypacker.create()
+  packer <- mx.nd.arraypacker()
   X$reset()
   while (X$iter.next()) {
     dlist = X$value()
@@ -354,13 +356,15 @@ predict.MXFeedForwardModel <- function(model, X, ctx=NULL, array.batch.size=128)
     oshape <- dim(out.pred)
     packer$push(mx.nd.slice(out.pred, 0, oshape[[1]] - padded))
   }
+  X$reset()
   return(packer$get())
 }
 
 #' Load model checkpoint from file.
 #'
 #' @param prefix string prefix of the model name
 #' @param iteration integer Iteration number of model we would like to load.
+#'
 #' @export
 mx.model.load <- function(prefix, iteration) {
   symbol <- mx.symbol.load(paste0(prefix, "-symbol.json"))
@@ -398,6 +402,7 @@ mx.model.load <- function(prefix, iteration) {
 #' @param model The feedforward model to be saved.
 #' @param prefix string prefix of the model name
 #' @param iteration integer Iteration number of model we would like to load.
+#'
 #' @export
 mx.model.save <- function(model, prefix, iteration) {
   arg.params <- model$arg.params