H2O xgboost mojo预测警告

Question

我试图使用H2O中的XGBoost mojo对象从新数据中生成预测。但是，当我预测时，会显示各种消息，发出警告

Feb 26, 2019 12:43:47 PM ml.dmlc.xgboost4j.java.NativeLibrary extractAndLoad
WARNING: Cannot load library from path lib/linux_64/libxgboost4j_gpu.so
Feb 26, 2019 12:43:47 PM ml.dmlc.xgboost4j.java.NativeLibrary extractAndLoad
WARNING: Cannot load library from path lib/libxgboost4j_gpu.so
Feb 26, 2019 12:43:47 PM ml.dmlc.xgboost4j.java.NativeLibrary doLoad
WARNING: Failed to load library from both native path and jar!
Feb 26, 2019 12:43:47 PM ml.dmlc.xgboost4j.java.NativeLibraryLoaderChain loadNativeLibs
INFO: Cannot load library: xgboost4j_gpu (lib/linux_64/libxgboost4j_gpu.so)
Feb 26, 2019 12:43:47 PM ml.dmlc.xgboost4j.java.NativeLibrary extractAndLoad
INFO: Loaded library from lib/linux_64/libxgboost4j_omp.so (/tmp/libxgboost4j_omp7945713229272382570.so)

       predict          setosa      versicolor       virginica
1       setosa 0.9961976408958 0.0030118888244 0.0007904054946
2       setosa 0.9963765740395 0.0026796606835 0.0009437160916
3       setosa 0.9963235855103 0.0028859297745 0.0007905053790
4       setosa 0.9963260293007 0.0028859369922 0.0007880008779
5       setosa 0.9961976408958 0.0030118888244 0.0007904054946

这是一个基本可重复的例子：

library(tidyverse)
library(h2o)

h2o.init(nthreads = -1, max_mem_size = '5g') # All available cores

data(iris)
iris.hex <- as.h2o(iris, destination_frame = "iris.hex")

iris.gbm <- h2o.xgboost(y = 5, x = 1:4, training_frame = iris.hex, ntrees = 100,
                        max_depth = 3,
                        learn_rate = 0.2,
                        distribution= "AUTO")

h2o.download_mojo(iris.gbm, "Mojo_models/", get_genmodel_jar = T)
h2o.shutdown()

我尝试执行预测时会出现警告消息

h2o.mojo_predict_df(iris, mojo_zip_path = "Mojo_models/XGBoost_model_R_1551184956713_1.zip", genmodel_jar_path = "Mojo_models/h2o-genmodel.jar", 
                    java_option =  '-Xmx1g -XX:ReservedCodeCacheSize=256m')

这是我应该担心的吗？如果不是，为什么消息会发生，我可以以某种方式抑制它们吗？

相关会议信息：

R version 3.4.2 (2017-09-28)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 14.04.4 LTS

Matrix products: default
BLAS: /usr/lib/libblas/libblas.so.3.0
LAPACK: /usr/lib/lapack/liblapack.so.3.0

other attached packages:
 [1] h2o_3.22.1.1    forcats_0.3.0   stringr_1.3.1   dplyr_0.7.8     purrr_0.3.0     readr_1.2.1     tidyr_0.8.2     tibble_1.4.2    ggplot2_3.1.0   tidyverse_1.2.1

Answer 1

H2O正在尝试并且未能加载GPU版本XGBoost。然后加载CPU版本。除非您认为GPU版本应该加载，否则这不是值得担心的事情。