Version 1.1.5 (#69)

bugfix for LRFinder with a device function
OHDSI · Apr 24, 2023 · cfb1bfa · cfb1bfa
1 parent f26382e
commit cfb1bfa
Show file tree

Hide file tree

Showing 4 changed files with 28 additions and 2 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: DeepPatientLevelPrediction
 Type: Package
 Title: Deep Learning For Patient Level Prediction Using Data In The OMOP Common Data Model
-Version: 1.1.4
+Version: 1.1.5
 Date: 18-04-2023
 Authors@R: c(
     person("Egill", "Fridgeirsson", email = "e.fridgeirsson@erasmusmc.nl", role = c("aut", "cre")),

diff --git a/NEWS.md b/NEWS.md
@@ -1,3 +1,7 @@
+DeepPatientLevelPrediction 1.1.5
+======================
+  - Fix bug where device function was not working for LRFinder
+
 DeepPatientLevelPrediction 1.1.4
 ======================
  - Remove torchopt dependancy since adamw is now in torch

diff --git a/R/LRFinder.R b/R/LRFinder.R
@@ -78,7 +78,7 @@ lrFinder <- function(dataset, modelType, modelParams, estimatorSettings,
       optimizer$zero_grad()
 
       batch <- dataset[sample(batchIndex, estimatorSettings$batchSize)]
-      batch <- batchToDevice(batch, device=estimatorSettings$device)
+      batch <- batchToDevice(batch, device=device)
 
       output <- model(batch$batch)
 

diff --git a/tests/testthat/test-LRFinder.R b/tests/testthat/test-LRFinder.R
@@ -39,6 +39,28 @@ test_that("LR finder works", {
 
   expect_true(lr<=10.0)
   expect_true(lr>=3e-4)
+})
+
+test_that("LR finder works with device specified by a function", {
+
+  deviceFun <- function(){
+    dev = "cpu"
+  }
+  lr <- lrFinder(dataset, modelType = ResNet, modelParams = list(catFeatures=dataset$numCatFeatures(),
+                                                                 numFeatures=dataset$numNumFeatures(),
+                                                                 sizeEmbedding=8,
+                                                                 sizeHidden=16,
+                                                                 numLayers=1,
+                                                                 hiddenFactor=1),
+                 estimatorSettings = setEstimator(batchSize=32,
+                                                  seed = 42,
+                                                  device = deviceFun),
+                 minLR = 3e-4,
+                 maxLR = 10.0,
+                 numLR = 20,
+                 divergenceThreshold = 1.1)
+  expect_true(lr<=10.0)
+  expect_true(lr>=3e-4)
 
 
 })