diff --git a/README.md b/README.md
index 236dae5..e4e9057 100644
--- a/README.md
+++ b/README.md
@@ -71,7 +71,7 @@ python eval_retrieval.py --bert_model bert-base-uncased --from_pretrained save/R
 2: To test on held out validation split, use the following command: 
 
 ```
-
+python eval_tasks.py --bert_model bert-base-uncased --from_pretrained save/VQA_bert_base_6layer_6conect-pretrained/pytorch_model_19.bin --config_file config/bert_base_6layer_6conect.json --task 0 --split minval
 ```
 
 ### VCR
@@ -81,13 +81,13 @@ python eval_retrieval.py --bert_model bert-base-uncased --from_pretrained save/R
 2: To test on VCR Q->A
 
 ```
-
+python eval_tasks.py --bert_model bert-base-uncased --from_pretrained save/VCR_Q-A-VCR_QA-R_bert_base_6layer_6conect-pretrained/pytorch_model_19.bin --config_file config/bert_base_6layer_6conect.json --task 1 --split val
 ```
 
 3: To test on VCR QA->R
 
 ```
-
+python eval_tasks.py --bert_model bert-base-uncased --from_pretrained save/VCR_Q-A-VCR_QA-R_bert_base_6layer_6conect-pretrained/pytorch_model_19.bin --config_file config/bert_base_6layer_6conect.json --task 2 --split val
 ```
 
 ### RefCOCO+
diff --git a/vilbert/datasets/vcr_dataset.py b/vilbert/datasets/vcr_dataset.py
index 0c7dfb4..9b62232 100644
--- a/vilbert/datasets/vcr_dataset.py
+++ b/vilbert/datasets/vcr_dataset.py
@@ -122,6 +122,9 @@ def __init__(
                 if row[1] != 'name':
                     self._names.append(row[1])
 
+        if not os.path.exists(os.path.join(dataroot, "cache")):
+            os.makedirs(os.path.join(dataroot, "cache"))
+
         # cache file path data/cache/train_ques
         cache_path = "data/VCR/cache/" + split + '_' + task + "_" + str(max_seq_length) + "_" + str(max_region_num) + "_vcr.pkl"
         if not os.path.exists(cache_path):