version 1.1.4 (#67)

Adds device input as a function to estimator

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: DeepPatientLevelPrediction
 Type: Package
 Title: Deep Learning For Patient Level Prediction Using Data In The OMOP Common Data Model
-Version: 1.1.3
-Date: 15-12-2022
+Version: 1.1.4
+Date: 18-04-2023
 Authors@R: c(
     person("Egill", "Fridgeirsson", email = "[email protected]", role = c("aut", "cre")),
     person("Jenna", "Reps", email = "[email protected]", role = c("aut")), 
@@ -24,8 +24,7 @@ Imports:
     ParallelLogger (>= 2.0.0),
     PatientLevelPrediction (>= 6.0.4),
     rlang,
-    torch (>= 0.9.0),
-    torchopt,
+    torch (>= 0.10.0),
     withr
 Suggests:
     devtools,

NEWS.md

@@ -1,3 +1,9 @@
+DeepPatientLevelPrediction 1.1.4
+======================
+ - Remove torchopt dependancy since adamw is now in torch
+ - Update torch dependency to >=0.10.0
+ - Allow device to be a function that resolves during Estimator initialization
+
 DeepPatientLevelPrediction 1.1.3
 ======================
 - Fix actions after torch updated to v0.10 (#65)

R/Estimator-class.R

@@ -34,7 +34,12 @@ Estimator <- R6::R6Class(
                           modelParameters,
                           estimatorSettings) {
       self$seed <- estimatorSettings$seed
-      self$device <- estimatorSettings$device
+      if (is.function(estimatorSettings$device)) {
+        device <- estimatorSettings$device()
+      } else {
+        device <- estimatorSettings$device
+      }
+      self$device <- device
       torch::torch_manual_seed(seed=self$seed)
       self$model <- do.call(modelType, modelParameters)
       self$modelParameters <- modelParameters

R/Estimator.R

@@ -26,7 +26,8 @@
 #' @param weightDecay what weight_decay to use
 #' @param batchSize batchSize to use
 #' @param epochs  how many epochs to train for
-#' @param device  what device to train on
+#' @param device  what device to train on, can be a string or a function to that evaluates
+#' to the device during runtime
 #' @param optimizer which optimizer to use
 #' @param scheduler which learning rate scheduler to use
 #' @param criterion loss function to use
@@ -41,7 +42,7 @@ setEstimator <- function(learningRate='auto',
                          batchSize = 512,
                          epochs = 30,
                          device='cpu',
-                         optimizer = torchopt::optim_adamw,
+                         optimizer = torch::optim_adamw,
                          scheduler = list(fun=torch::lr_reduce_on_plateau,
                                           params=list(patience=1)),
                          criterion = torch::nn_bce_with_logits_loss,

tests/testthat/test-Estimator.R

@@ -298,4 +298,42 @@ test_that("setEstimator with paramsToTune is correctly added to hyperparameters"
   expect_equal(estimatorSettings2$learningRate, 1e-3)
   expect_equal(as.character(estimatorSettings2$metric), "auprc")
   expect_equal(estimatorSettings2$earlyStopping$params$patience, 10)
-})
+})
+
+test_that("device as a function argument works", {
+  getDevice <- function() {
+    dev <- Sys.getenv("testDeepPLPDevice") 
+    if (dev == ""){
+      dev = "cpu"
+    } else{
+      dev
+    }
+  }
+
+  estimatorSettings <- setEstimator(device=getDevice)
+
+  model <- setDefaultResNet(estimatorSettings = estimatorSettings) 
+  model$param[[1]]$catFeatures <- 10
+
+  estimator <- Estimator$new(modelType="ResNet",
+                             modelParameters = model$param[[1]],
+                             estimatorSettings = estimatorSettings)
+
+  expect_equal(estimator$device, "cpu")
+
+  Sys.setenv("testDeepPLPDevice" = "meta")
+
+  estimatorSettings <- setEstimator(device=getDevice)
+
+  model <- setDefaultResNet(estimatorSettings = estimatorSettings) 
+  model$param[[1]]$catFeatures <- 10
+
+  estimator <- Estimator$new(modelType="ResNet",
+                             modelParameters = model$param[[1]],
+                             estimatorSettings = estimatorSettings)
+
+  expect_equal(estimator$device, "meta")
+
+  Sys.unsetenv("testDeepPLPDevice")
+
+  })

tests/testthat/test-LRFinder.R

@@ -3,7 +3,7 @@ test_that("LR scheduler that changes per batch works", {
   model <- ResNet(catFeatures = 10, numFeatures = 1,
                   sizeEmbedding = 32, sizeHidden = 64,
                   numLayers = 1, hiddenFactor = 1)
-  optimizer <- torchopt::optim_adamw(model$parameters, lr=1e-7)
+  optimizer <- torch::optim_adamw(model$parameters, lr=1e-7)
 
   scheduler <- lrPerBatch(optimizer,
                           startLR = 1e-7,