diff --git a/bertserini/experiments/eval/evaluate_v1_cmrc.py b/bertserini/experiments/eval/evaluate_v1_cmrc.py
index 0f3ff32..4556d15 100755
--- a/bertserini/experiments/eval/evaluate_v1_cmrc.py
+++ b/bertserini/experiments/eval/evaluate_v1_cmrc.py
@@ -17,6 +17,9 @@
 #from utils import init_logger
 #logger = init_logger("evaluation")
 
+#install punkt corpus
+nltk.download('punkt')
+
 # split Chinese with English
 def mixed_segmentation(in_str, rm_punc=False):
 	in_str = str(in_str).lower().strip()
diff --git a/docs/experiments-cmrc.md b/docs/experiments-cmrc.md
index ab63f13..c551260 100644
--- a/docs/experiments-cmrc.md
+++ b/docs/experiments-cmrc.md
@@ -58,7 +58,7 @@ python -m bertserini.experiments.inference --dataset_path data/cmrc2018_dev_squa
 
 ```
 mkdir temp
-pyhton -m bertserini.experiments.evaluate --eval_data data/cmrc2018_dev.json \
+python -m bertserini.experiments.evaluate --eval_data data/cmrc2018_dev.json \
                                           --search_file prediction/cmrc2018_pred.json \
                                           --output_path temp \
                                           --dataset cmrc
@@ -70,4 +70,4 @@ Expected results:
 ```
 ## rsvp-ai/bertserini-bert-base-cmrc, this is bert-base-chinese finetuned on the chinese reading comprehension dataset(CMRC)
 (0.5, {'f1_score': 68.0033167812909, 'exact_match': 51.164958061509786, 'total_count': 3219, 'skip_count': 1})
-```
\ No newline at end of file
+```
diff --git a/docs/experiments-squad.md b/docs/experiments-squad.md
index 105b67d..65a275b 100644
--- a/docs/experiments-squad.md
+++ b/docs/experiments-squad.md
@@ -63,7 +63,7 @@ python -m bertserini.experiments.inference --dataset_path data/dev-v1.1.json \
 
 ```
 mkdir temp
-pyhton -m bertserini.experiments.evaluate --eval_data data/dev-v1.1.json \
+python -m bertserini.experiments.evaluate --eval_data data/dev-v1.1.json \
                                           --search_file prediction/squad_bert_large_pred.json \
                                           --output_path temp \
                                           --dataset squad
@@ -76,4 +76,4 @@ Expected results:
 
 ## rsvp-ai/bertserini-bert-base-squad, this is finetuned based on bert-base-uncased
 (0.5, {'exact_match': 40.179754020813625, 'f1': 47.828056659017584, 'recall': 49.517951036176, 'precision': 48.3495034100538, 'cover': 45.50614947965941, 'overlap': 56.20624408703879})
-```
\ No newline at end of file
+```