fix black in pre-commit (#1940)

test/srt/models/test_generation_models.py

@@ -56,7 +56,7 @@ ALL_OTHER_MODELS = [
     ModelCase("HuggingFaceTB/SmolLM-135M-Instruct", skip_long_prompt=True),
     ModelCase("allenai/OLMo-1B-0724-hf", decode_tolerance=8e-2, skip_long_prompt=True),
     ModelCase("THUDM/glm-4-9b-chat"),
-    ModelCase("openai-community/gpt2")
+    ModelCase("openai-community/gpt2"),
 ]
 
 TORCH_DTYPES = [torch.float16]

test/srt/test_openai_server.py

@@ -3,6 +3,7 @@ python3 -m unittest test_openai_server.TestOpenAIServer.test_batch
 python3 -m unittest test_openai_server.TestOpenAIServer.test_completion
 
 """
+
 import json
 import time
 import unittest

test/srt/test_skip_tokenizer_init.py

@@ -1,6 +1,7 @@
 """
 python3 -m unittest test_skip_tokenizer_init.TestSkipTokenizerInit.test_parallel_sample
 """
+
 import json
 import unittest
 

test/srt/test_srt_engine.py

@@ -110,7 +110,6 @@ class TestSRTEngine(unittest.TestCase):
     def test_5_prompt_input_ids_consistency(self):
         prompt = "The capital of UK is"
 
-
         model_path = DEFAULT_MODEL_NAME_FOR_TEST
         engine = sgl.Engine(model_path=model_path, random_seed=42, log_level="error")
         sampling_params = {"temperature": 0, "max_new_tokens": 8}
@@ -118,7 +117,9 @@ class TestSRTEngine(unittest.TestCase):
 
         tokenizer = get_tokenizer(model_path)
         token_ids = tokenizer.encode(prompt)
-        out2 = engine.generate(input_ids=token_ids, sampling_params=sampling_params)["text"]
+        out2 = engine.generate(input_ids=token_ids, sampling_params=sampling_params)[
+            "text"
+        ]
 
         engine.shutdown()