diff --git a/downloads-generation/models_class1_allele_specific_single/README.md b/downloads-generation/models_class1_allele_specific_single/README.md
index 12046a005e62d761d5cbccf7b6f5835d7fc48c3b..0003e3cd699e1122401f2797060815a7ab768d00 100644
--- a/downloads-generation/models_class1_allele_specific_single/README.md
+++ b/downloads-generation/models_class1_allele_specific_single/README.md
@@ -2,15 +2,13 @@
 
 This download contains trained MHC Class I allele-specific MHCflurry models. The training data used is in the [data_combined_iedb_kim2014](../data_combined_iedb_kim2014) MHCflurry download. We first select network hyperparameters for each allele individually using cross validation over the models enumerated in [models.py](models.py). The best hyperparameter settings are selected via average of AUC (at 500nm), F1, and Kendall's Tau over the training folds. We then train the production models over the full training set using the selected hyperparameters.
 
-The training script supports multi-node parallel execution using the [dask-distributed](https://distributed.readthedocs.io/en/latest/) library. To enable this, pass the IP and port of the dask scheduler to the training script with the '--dask-scheduler' option. The GENERATE.sh script passes all arguments to the training script so you can just give it as an argument to GENERATE.sh.
+The training script supports multi-node parallel execution using the [kubeface](https://github.com/hammerlab/kubeface) librarie.
 
-We run dask distributed on Google Container Engine using Kubernetes as described [here](https://github.com/hammerlab/dask-distributed-on-kubernetes).
+To use kubeface, you should make a google storage bucket and pass it below with the --storage-prefix argument. 
 
 To generate this download we run:
 
 ```
-# If you are running dask distributed using our kubernetes config, you can use the DASK_IP one liner below.
-# Otherwise, just set it to the IP of the dask scheduler.
 ./GENERATE.sh \
     --cv-folds-per-task 10 \
     --backend kubernetes \
diff --git a/mhcflurry/class1_allele_specific/cross_validation.py b/mhcflurry/class1_allele_specific/cross_validation.py
index 184bae580ea4ed0c87efd9f0f9b9720a7a66dee1..1bf0038fdde8f9ab2a6808693bc18d0e93fdba75 100644
--- a/mhcflurry/class1_allele_specific/cross_validation.py
+++ b/mhcflurry/class1_allele_specific/cross_validation.py
@@ -187,6 +187,9 @@ def cross_validation_folds(
             lambda kwargs: impute_and_select_allele(**kwargs),
             imputation_args)
 
+        # Here _replace is a method on named tuples that returns a new named
+        # tuple with the specified key set to the given value and all other key/
+        # values the same as the original.
         return [
             result_fold._replace(imputed_train=imputation_result)
             for (result_fold, imputation_result) in zip(
diff --git a/test/test_class1_allele_specific_cv_and_train_command.py b/test/test_class1_allele_specific_cv_and_train_command.py
index 99d1a991139914da6d823ae968525d1303081709..bcc71db1e63e8f35e0e5be41f907943c1d7d650a 100644
--- a/test/test_class1_allele_specific_cv_and_train_command.py
+++ b/test/test_class1_allele_specific_cv_and_train_command.py
@@ -69,6 +69,9 @@ def test_small_run():
         "--num-local-threads", "1",
     ]
     if KUBEFACE_INSTALLED:
+        # If kubeface is installed, then this command will by default use it.
+        # In that case, we want to have the kubeface storage written to a
+        # local file and not assume the existence of a google storage bucket.
         args.extend(["--storage-prefix", "/tmp/"])
     print("Running cv_and_train_command with args: %s " % str(args))