diff --git a/tests/README.md b/tests/README.md
index e5c529bd..3b2b77fb 100644
--- a/tests/README.md
+++ b/tests/README.md
@@ -1,42 +1,53 @@
-# Unit tests
+# Unit Tests
 
 [![codecov](https://codecov.io/gh/huseinzol05/malaya/branch/master/graph/badge.svg?token=Mto5hHr8da)](https://codecov.io/gh/huseinzol05/malaya)
 
-## how-to
+This folder contains the unit tests for malaya. We use `pytest` as our testing framework.
 
-1. Install pytest,
+## Preqrequisites
+> Note: You should use Python 3.6 and above with pip3. Make sure to also use a virtual environment!
+Install pytest, pytest-cov, pytest-codecov, and gitpython.
 
 ```bash
 pip3 install pytest pytest-cov pytest-codecov gitpython
 ```
 
-2. Run pytest,
+Make sure tensorflow is installed, if some tests fail you might need to install the extra libraries in requirements.txt.
+
+```bash
+pip3 install -r requirements.txt
+```
+
+## Running Tests
+
+**Run all tests**
 
 ```bash
 pytest tests --cov --cov-report term --cov-report html
 ```
 
-Or run failed tests only,
+**Run failed tests only**
 
 ```bash
 pytest tests --cov --cov-report term --cov-report html --last-failed
 ```
 
-Or run for specific py file,
+**Run a specific test file**
 
 ```bash
-pytest tests/test_emotion.py --cov --cov-report term --cov-report html
+pytest tests/test_name.py --cov --cov-report term --cov-report html
 ```
 
-Or run for specific function,
+**Run a test for a specific function**
 
 ```bash
-pytest tests/test_emotion.py::test_multinomial --cov --cov-report term --cov-report html
+pytest tests/test_name.py::test_multinomial --cov --cov-report term --cov-report html
 ```
 
-3. Upload to CodeCov, https://app.codecov.io/gh/huseinzol05/malaya
+## Updating Code Coverage
+Once you have run all the tests upload the generated HTML coverage report to our CodeCov at https://app.codecov.io/gh/huseinzol05/malaya.
 
-```
+```bash
 CODECOV_TOKEN=
 pytest tests --cov --codecov --codecov-token=$CODECOV_TOKEN
 ```
\ No newline at end of file
diff --git a/tests/requirements.txt b/tests/requirements.txt
new file mode 100644
index 00000000..2ff8ae4f
--- /dev/null
+++ b/tests/requirements.txt
@@ -0,0 +1,2 @@
+youtokentome
+fasttext
\ No newline at end of file
diff --git a/tests/tests/test_abstractive_summarization.py b/tests/tests/test_abstractive_summarization.py
index 1a73c766..a04f33d2 100644
--- a/tests/tests/test_abstractive_summarization.py
+++ b/tests/tests/test_abstractive_summarization.py
@@ -48,7 +48,7 @@ def cleaning(string):
 
 def test_transformer():
     models = malaya.summarization.abstractive.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.summarization.abstractive.transformer(model=m)
         print(model.greedy_decoder([string]))
diff --git a/tests/tests/test_constituency.py b/tests/tests/test_constituency.py
index 1049f29a..6a0488e8 100644
--- a/tests/tests/test_constituency.py
+++ b/tests/tests/test_constituency.py
@@ -11,10 +11,9 @@
 
 string = 'Dr Mahathir menasihati mereka supaya berhenti berehat dan tidur sebentar sekiranya mengantuk ketika memandu.'
 
-
 def test_transformer():
     models = malaya.constituency.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.constituency.transformer(model=m, gpu_limit=0.3)
         print(model.parse_tree(string))
diff --git a/tests/tests/test_dependency.py b/tests/tests/test_dependency.py
index 833c81e0..c0cc517c 100644
--- a/tests/tests/test_dependency.py
+++ b/tests/tests/test_dependency.py
@@ -14,7 +14,7 @@
 
 def test_transformer_v2():
     models = malaya.dependency.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.dependency.transformer(model=m, gpu_limit=0.3)
         d_object, tagging, indexing = model.predict(string)
@@ -24,7 +24,7 @@ def test_transformer_v2():
 
 def test_transformer_v1():
     models = malaya.dependency.available_transformer(version='v1')
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.dependency.transformer(version='v1', model=m, gpu_limit=0.3)
         d_object, tagging, indexing = model.predict(string)
diff --git a/tests/tests/test_emotion.py b/tests/tests/test_emotion.py
index a459effd..ed68da46 100644
--- a/tests/tests/test_emotion.py
+++ b/tests/tests/test_emotion.py
@@ -19,7 +19,7 @@ def test_multinomial():
 
 def test_transformer():
     models = malaya.emotion.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.emotion.transformer(model=m, gpu_limit=0.3)
         print(model.predict_proba([text]))
diff --git a/tests/tests/test_entity.py b/tests/tests/test_entity.py
index b01674d4..29e04b01 100644
--- a/tests/tests/test_entity.py
+++ b/tests/tests/test_entity.py
@@ -15,7 +15,7 @@
 
 def test_transformer():
     models = malaya.entity.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.entity.transformer(model=m)
         print(model.predict(string))
@@ -27,7 +27,7 @@ def test_transformer():
 
 def test_transformer_ontonotes5():
     models = malaya.entity.available_transformer_ontonotes5()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.entity.transformer_ontonotes5(model=m)
         print(model.predict(string))
diff --git a/tests/tests/test_paraphrase.py b/tests/tests/test_paraphrase.py
index 9132629e..e02bb6d5 100644
--- a/tests/tests/test_paraphrase.py
+++ b/tests/tests/test_paraphrase.py
@@ -14,7 +14,7 @@
 
 def test_transformer():
     models = malaya.paraphrase.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.paraphrase.transformer(model=m)
         print(model.greedy_decoder([string]))
diff --git a/tests/tests/test_pos.py b/tests/tests/test_pos.py
index 9c855664..fda0a120 100644
--- a/tests/tests/test_pos.py
+++ b/tests/tests/test_pos.py
@@ -15,7 +15,7 @@
 
 def test_transformer():
     models = malaya.pos.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.pos.transformer(model=m)
         print(model.predict(string))
diff --git a/tests/tests/test_relevancy.py b/tests/tests/test_relevancy.py
index feab7a08..61880eff 100644
--- a/tests/tests/test_relevancy.py
+++ b/tests/tests/test_relevancy.py
@@ -14,7 +14,7 @@
 
 def test_transformer():
     models = malaya.relevancy.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.relevancy.transformer(model=m, gpu_limit=0.3)
         print(model.predict_proba([text]))
diff --git a/tests/tests/test_segmentation.py b/tests/tests/test_segmentation.py
index bd34a930..992dc806 100644
--- a/tests/tests/test_segmentation.py
+++ b/tests/tests/test_segmentation.py
@@ -14,7 +14,7 @@
 
 def test_transformer():
     models = malaya.segmentation.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.segmentation.transformer(model=m)
         print(model.greedy_decoder([string1]))
diff --git a/tests/tests/test_sentiment.py b/tests/tests/test_sentiment.py
index 87713bd2..d72da7a2 100644
--- a/tests/tests/test_sentiment.py
+++ b/tests/tests/test_sentiment.py
@@ -19,7 +19,7 @@ def test_multinomial():
 
 def test_transformer():
     models = malaya.sentiment.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.sentiment.transformer(model=m, gpu_limit=0.3)
         print(model.predict_proba([text]))
diff --git a/tests/tests/test_toxicity.py b/tests/tests/test_toxicity.py
index f0d0cf98..2832ca0f 100644
--- a/tests/tests/test_toxicity.py
+++ b/tests/tests/test_toxicity.py
@@ -19,7 +19,7 @@ def test_multinomial():
 
 def test_transformer():
     models = malaya.toxicity.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.toxicity.transformer(model=m, gpu_limit=0.3)
         print(model.predict_proba([text]))
diff --git a/tests/tests/test_translation_en_ms.py b/tests/tests/test_translation_en_ms.py
index e41b29c3..55e62fcf 100644
--- a/tests/tests/test_translation_en_ms.py
+++ b/tests/tests/test_translation_en_ms.py
@@ -14,7 +14,7 @@
 
 def test_transformer():
     models = malaya.translation.en_ms.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.translation.en_ms.transformer(model=m, gpu_limit=0.3)
         print(model.greedy_decoder([random_string1]))
diff --git a/tests/tests/test_translation_ms_en.py b/tests/tests/test_translation_ms_en.py
index c3a79cbd..ff8b91b7 100644
--- a/tests/tests/test_translation_ms_en.py
+++ b/tests/tests/test_translation_ms_en.py
@@ -14,7 +14,7 @@
 
 def test_transformer():
     models = malaya.translation.ms_en.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.translation.ms_en.transformer(model=m, gpu_limit=0.3)
         print(model.greedy_decoder([random_string1]))
diff --git a/tests/tests/test_true_case.py b/tests/tests/test_true_case.py
index 934f508f..2c0708c4 100644
--- a/tests/tests/test_true_case.py
+++ b/tests/tests/test_true_case.py
@@ -14,7 +14,7 @@
 
 def test_transformer():
     models = malaya.true_case.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.true_case.transformer(model=m)
         print(model.greedy_decoder([string1]))
diff --git a/tests/tests/test_zeroshot.py b/tests/tests/test_zeroshot.py
index 3975a248..0aad4039 100644
--- a/tests/tests/test_zeroshot.py
+++ b/tests/tests/test_zeroshot.py
@@ -20,7 +20,7 @@
 
 def test_transformer():
     models = malaya.zero_shot.classification.available_transformer()
-    for m in models.index:
+    for m in models.keys():
         print(m)
         model = malaya.zero_shot.classification.transformer(model=m, gpu_limit=0.3)
         model.predict_proba([string], labels=['najib razak', 'mahathir', 'kerajaan', 'PRU', 'anarki'])