Merge pull request #2399 from axsaucedo/2065_servers_pinned_versions

Updated pinned versions of prepackaged servers

components/routers/case_study/models/xgb_model/requirements.txt

@@ -1,4 +1,2 @@
-scikit-learn==0.20.1
-numpy==1.15.1
-scipy==1.1.0
-xgboost==0.81
+numpy >= 1.8.2
+xgboost==1.2.0

doc/source/servers/sklearn.md

@@ -5,7 +5,10 @@ If you have a trained SKLearn model saved as a pickle you can deploy it simply u
 Pre-requisites:
 
   * The model pickle must be saved using joblib and presently be named `model.joblib`
-  * We presently use sklearn version 0.23.2. Your pickled model must be compatible with this version
+  * Installed dependencies (may not work if versions don't match):
+      + scikit-learn == 0.23.2
+      + joblib == 0.16.0
+      + numpy >= 1.8.2
 
 An example for a saved Iris prediction model:
 
@@ -105,4 +108,4 @@ If you wish the server image for the sklearn server to be globally changed you c
     rest:
       defaultImageVersion: "1.2.3"
       image: seldonio/sklearnserver_rest
-```
+```

doc/source/servers/xgboost.md

@@ -4,10 +4,12 @@ If you have a trained XGBoost model saved you can deploy it simply using Seldon'
 
 Prequisites:
 
-  * Use xgboost v0.82
   * The model must be named `model.bst`
   * You must save your model using `bst.save_model(file_path)`
   * The model is loaded with `xgb.Booster(model_file=model_file)`
+  * Dependencies (otherwise it may not work):
+      + numpy == 1.15.4
+      + xgboost == 1.2.0
 
 An example for a saved Iris prediction model:
 

servers/sklearnserver/sklearnserver/requirements.txt

@@ -1,5 +1,5 @@
 seldon_core
 scikit-learn == 0.23.2
 numpy >= 1.8.2
-joblib >= 0.13.0
+joblib == 0.16.0