fix pylint

Signed-off-by: Sukriti-Sharma4 <[email protected]>
foundation-model-stack · May 24, 2024 · 45827ce · 45827ce
1 parent 2f70e34
commit 45827ce
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 9 deletions.
diff --git a/tests/utils/test_data_utils.py b/tests/utils/test_data_utils.py
@@ -30,7 +30,9 @@ def test_formatting_function():
     json_dataset = datasets.load_dataset("json", data_files=TWITTER_COMPLAINTS_DATA)
     template = "### Input: {{Tweet text}} \n\n ### Response: {{text_label}}"
     # First response from the data file that is read.
-    expected_response = "### Input: @HMRCcustomers No this is my first job \n\n ### Response: no complaint"
+    expected_response = (
+        "### Input: No this is my first job \n\n ### Response: no complaint"
+    )
     formatted_dataset, dataset_text_field = data_utils.formatting_function(
         json_dataset, template
     )
@@ -43,7 +45,9 @@ def test_formatting_function_adds_eos_token():
     json_dataset = datasets.load_dataset("json", data_files=TWITTER_COMPLAINTS_DATA)
     template = "### Input: {{Tweet text}} \n\n ### Response: {{text_label}}"
     # First response from the data file that is read.
-    expected_response = "### Input: @HMRCcustomers No this is my first job \n\n ### Response: no complaintEOS"
+    expected_response = (
+        "### Input: No this is my first job \n\n ### Response: no complaintEOS"
+    )
     formatted_dataset, dataset_text_field = data_utils.formatting_function(
         json_dataset, template, "EOS"
     )

diff --git a/tuning/config/configs.py b/tuning/config/configs.py
@@ -54,7 +54,8 @@ class DataArguments:
         default=None,
         metadata={
             "help": "Training dataset text field containing single sequence. \
-                                        Either the dataset_text_field or data_formatter_template need to be supplied."
+                    Either the dataset_text_field \
+                    or data_formatter_template need to be supplied."
         },
     )
     validation_data_path: str = field(
@@ -64,9 +65,11 @@ class DataArguments:
     data_formatter_template: str = field(
         default=None,
         metadata={
-            "help": "formatter template to format a single sequence from each instance in JSONL files. \
-                                        Keys of JSON can be referred to as {{key}} in template. Either the dataset_text_field \
-                                        or data_formatter_template needs to be supplied."
+            "help": "formatter template to format a single sequence \
+                         from each instance in JSONL files. \
+                         Keys of JSON can be referred to as {{key}} in template. \
+                         Either the dataset_text_field \
+                         or data_formatter_template needs to be supplied."
         },
     )
 

diff --git a/tuning/utils/data_utils.py b/tuning/utils/data_utils.py
@@ -1,9 +1,6 @@
 # Standard
 import re
 
-# Third Party
-from datasets import Dataset
-
 
 def formatting_function(dataset, template, eos_token=""):
     """Function to format datasets with Alpaca style / other templates.