diff --git a/README.md b/README.md
index 0648fc3..ad1de52 100644
--- a/README.md
+++ b/README.md
@@ -1,15 +1,22 @@
 Train a GT4HistOCR Calamari model
 =================================
 
-`train.sh` trains a Calamari model based on GT4HistOCR. Or rather 5 using
-cross-validation to use for confidence voting.
-
-Requires Calamari 0.3.5.
-
-`train.sh` is able to download GT4HistOCR from the web if the `data` submodule
-is not available, that is if you're not a member of the Qurator team at SBB.
+`train.sh` trains a Calamari 1 model based on GT4HistOCR. Or rather 5 using
+cross-validation to use for confidence voting. This repository mainly
+serves as documentation of the provenance of the model published at
+https://qurator-data.de/calamari-models/, not as the definitive guide to
+training such a model.
 
 Trained models
 --------------
 For a finished model have a look here:
 https://qurator-data.de/calamari-models/
+
+Training your own model
+-----------------------
+If you really want to, you can use this script to train your own. It takes
+about 1 week on a Nvidia RTX 2080 GPU. Please use [requirements.txt](requirements.txt)
+in that case to setup a virtualenv.
+
+`train.sh` is able to download GT4HistOCR from the web if the `data` submodule
+is not available, that is if you're not a member of the Qurator team at SBB.