mlr3 version question for "paramSet and param_GL

[juinzhu]

Thanks anyone.
I have install the mlr3 and its related packages for built the survival machine learning models on the MAC, and then I have exported the leaners into the rdata files.
But when I install another packages and the mlr3 packages updated follow.
After then, I cannot open the leaners from the benchmark list. It turn out like that:
Blackboost_exp_all$model
LearnerSurvBlackBoost:surv.blackboost: Gradient Boosting

• Model: blackboost
  Error in .__ParamSet__values(self = self, private = private, super = super, :
  could not find function ".__ParamSet__values"

So, how can I do?

my mlr3 version were list blow
other attached packages:
[1] mlr3_0.18.0

loaded via a namespace (and not attached):
[1] paradox_1.0.0 colorspace_2.1-0 ggsignif_0.6.4
[4] class_7.3-22 modeltools_0.2-23 mclust_6.1
[7] set6_0.2.6 clue_0.3-65 rstudioapi_0.15.0
[10] distr6_1.8.4 ggpubr_0.6.0 listenv_0.9.1
[13] stats_4.2.3 mlr3tuning_0.20.0 flexmix_2.3-19
[16] mvtnorm_1.2-4 fansi_1.0.6 lubridate_1.9.3
[19] codetools_0.2-19 splines_4.2.3 mlr3learners_0.6.0
[22] robustbase_0.99-2 libcoin_1.0-10 knitr_1.45
[25] mlr3pipelines_0.5.0-9000 jsonlite_1.8.8 Formula_1.2-5
[28] RhpcBLASctl_0.23-42 broom_1.0.5 km.ci_0.5-6
[31] base_4.2.3 cluster_2.1.6 kernlab_0.9-32
[34] data.tree_1.1.0 DiagrammeR_1.0.11 readr_2.1.5
[37] compiler_4.2.3 randomForestSRC_3.2.3 backports_1.4.1
[40] fastmap_1.1.1 survex_1.2.0 Matrix_1.6-5
[43] cli_3.6.2 visNetwork_2.1.2 htmltools_0.5.7
[46] tools_4.2.3 partykit_1.2-20 ooplah_0.2.0
[49] gtable_0.3.4 glue_1.7.0 dplyr_1.1.4
[52] grDevices_4.2.3 Rcpp_1.0.12 mlr3hyperband_0.5.0
[55] carData_3.0-5 vctrs_0.6.5 fpc_2.2-11
[58] inum_1.0-5 xfun_0.42 stringr_1.5.1
[61] globals_0.16.3 timechange_0.3.0 mlr3verse_0.2.8
[64] lifecycle_1.0.4 rstatix_0.7.2 future_1.33.1
[67] DEoptimR_1.1-3 MASS_7.3-60.0.1 zoo_1.8-12
[70] scales_1.3.0 lgr_0.4.4 graphics_4.2.3
[73] mlr3filters_0.7.1 hms_1.1.3 parallel_4.2.3
[76] tidyverse_2.0.0 RColorBrewer_1.1-3 spacefillr_0.3.2
[79] mlr3tuningspaces_0.5.0 utils_4.2.3 gridExtra_2.3
[82] mlr3data_0.7.0 KMsurv_0.1-5 ggplot2_3.5.0
[85] stabs_0.6-4 datasets_4.2.3 rpart_4.1.23
[88] mlr3fselect_0.12.0 stringi_1.8.3 checkmate_2.3.1
[91] dictionar6_0.1.3 palmerpenguins_0.1.1 mlr3viz_0.8.0
[94] rlang_1.1.3 pkgconfig_2.0.3 prabclus_2.3-3
[97] lattice_0.22-6 purrr_1.0.2 htmlwidgets_1.6.4
[100] patchwork_1.2.0 mboost_2.9-9 tidyselect_1.2.1
[103] parallelly_1.37.1 magrittr_2.0.3 R6_2.5.1
[106] nnls_1.5 generics_0.1.3 mlr3mbo_0.2.2
[109] mlr3proba_0.6.0 pillar_1.9.0 param6_0.2.4
[112] withr_3.0.0 DALEX_2.4.3 survival_3.5-8
[115] abind_1.4-5 nnet_7.3-19 tibble_3.2.1
[118] mlr3cluster_0.1.9 crayon_1.5.2 car_3.1-2
[121] survMisc_0.5.6 uuid_1.2-0 utf8_1.2.4
[124] tzdb_0.4.0 grid_4.2.3 data.table_1.15.2
[127] methods_4.2.3 forcats_1.0.0 mlr3misc_0.14.0
[130] bbotk_0.8.0 digest_0.6.35 diptest_0.77-0
[133] xtable_1.8-4 tidyr_1.3.1 stats4_4.2.3
[136] munsell_0.5.0 survminer_0.4.9 quadprog_1.5-8
[139] mlr3extralearners_0.7.1

[juinzhu]

When I load the benchmark file from the Windows, with the mlr3 version 0.18.0, it turns out that
Coxnet_model_all = surv_exp_f(1,1)[[1]]
Warning messages:
1: Detected version mismatch: Learner 'surv.glmnet' has been trained with mlr3 version '0.17.2.9000', not matching currently installed version '0.18.0'
2: Multiple lambdas have been fit. Lambda will be set to 0.01 (see parameter 's').
But when I reinstall the mlr3 package for version 0.17.2.9000, it report the error for no version.

[mb706]

The .9000 in the version number 0.17.2.9000 indicates that the model was trained with a GitHub-version of mlr3. Simply reinstalling "0.17.2.9000" might potentially not be enough to solve the version conflict, since it might still be a different commit of mlr3 that you are installing.

[mb706]

May be an issue with paradox. Could you try to install the CRAN version of paradox (you currently have the github version) and see if that helps?