minor: Add basic editorconfig and pre-commit hooks to enforce style for whitespaces (#1926)

python/sglang/srt/layers/quantization/base_config.py

@@ -134,4 +134,4 @@ def method_has_implemented_embedding(
     class_embedding = inspect.getattr_static(method_class, "embedding", None)
 
     return (class_embedding is not None
-            and class_embedding is not base_embedding)
+            and class_embedding is not base_embedding)

python/sglang/srt/layers/vocab_parallel_embedding.py

@@ -311,7 +311,7 @@ class VocabParallelEmbedding(torch.nn.Module):
     def get_sharded_to_full_mapping(self) -> Optional[List[int]]:
         """Get a mapping that can be used to reindex the gathered
         logits for sampling.
-        
+
         During sampling, we gather logits from all ranks. The relationship
         of index->token_id will follow the same format as outlined in the class
         docstring. However, after the gather, we want to reindex the final
@@ -483,4 +483,4 @@ class ParallelLMHead(VocabParallelEmbedding):
 
     def forward(self, input_):
         del input_
-        raise RuntimeError("LMHead's weights should be used in the sampler.")
+        raise RuntimeError("LMHead's weights should be used in the sampler.")

python/sglang/srt/managers/scheduler.py

@@ -838,7 +838,7 @@ class Scheduler:
         time_per_output_tokens_iter: List[float] = []
 
         # Request stats
-        #   Decode 
+        #   Decode
         gen_throughput: float = 0.0
         #   Latency
         time_e2e_requests: List[float] = []
@@ -866,11 +866,11 @@ class Scheduler:
                     time_waiting_requests.append(req.queued_time - req.created_time)
                     num_prompt_tokens_requests.append(len(req.origin_input_ids))
                     num_generation_tokens_requests.append(len(req.output_ids))
-                    finished_reason_requests.append(                            
+                    finished_reason_requests.append(
                             req.finished_reason.to_json()
                             if req.finished_reason is not None
                             else None)
-    
+
         return Stats(
             new_seq=new_seq,
             num_running_req=num_running_req,

python/sglang/srt/managers/tokenizer_manager.py

@@ -384,7 +384,7 @@ class TokenizerManager:
             obj.load_format = self.server_args.load_format
 
         if not self.model_update_lock.locked():
-        
+
             async with self.model_update_lock:
                 # wait for the previous generation requests to finish
                 while len(self.rid_to_state) > 0:

python/sglang/srt/metrics/metrics_collector.py

@@ -151,7 +151,7 @@ class Metrics:
                 0.005, 0.01, 0.015, 0.02, 0.025, 0.03, 0.04, 0.05, 0.075, 0.1, 0.15, 0.2, 0.3, 0.4, 0.5, 0.75,
                 1.0, 2.5
             ])
-        
+
         # Request Stats
         #   Metadata
         self.num_prompt_tokens_requests = Histogram(
@@ -253,7 +253,7 @@ class PrometheusMetricsCollector(MetricsCollector):
                             stats.time_to_first_tokens_iter)
         self._log_histogram(self.metrics.histogram_time_per_output_token,
                             stats.time_per_output_tokens_iter)
-        
+
         # self._log_gauge(self.metrics.gpu_cache_usage_sys, stats.gpu_cache_usage_sys)
         self._log_gauge(self.metrics.num_running_sys, stats.num_running_req)
         self._log_gauge(self.metrics.num_waiting_sys, stats.num_waiting_req)
@@ -294,4 +294,4 @@ def build_1_2_5_buckets(max_value: int) -> List[int]:
                 buckets.append(value)
             else:
                 return buckets
-        exponent += 1
+        exponent += 1

python/sglang/srt/metrics/metrics_types.py

@@ -54,4 +54,4 @@ class Stats:
     num_prompt_tokens_iter: int = 0
     num_generation_tokens_iter: int = 0
     time_to_first_tokens_iter: List[float] = field(default_factory=list)
-    time_per_output_tokens_iter: List[float] = field(default_factory=list)
+    time_per_output_tokens_iter: List[float] = field(default_factory=list)

python/sglang/srt/mm_utils.py

@@ -17,7 +17,7 @@ limitations under the License.
 """
 Utilities for multi-modal models.
 
-This python file mainly contains utilities that were used in the 
+This python file mainly contains utilities that were used in the
 image processing logic of llava-next including operations such as
 anyres and anyres_max
 

python/sglang/srt/models/gpt2.py

@@ -136,7 +136,7 @@ class GPT2Block(nn.Module):
         layer_id: int,
         config: GPT2Config,
         cache_config = None,
-        
+
         quant_config: Optional[QuantizationConfig] = None,
         prefix: str = "",
     ):
@@ -284,4 +284,4 @@ class GPT2LMHeadModel(nn.Module):
                                     default_weight_loader)
             weight_loader(param, loaded_weight)
 
-EntryClass = GPT2LMHeadModel
+EntryClass = GPT2LMHeadModel

python/sglang/srt/models/qwen2_vl.py

@@ -57,27 +57,27 @@ logger = init_logger(__name__)
 
 class Qwen2VLImageInputs(TypedDict):
     pixel_values: torch.Tensor
-    """Shape: 
+    """Shape:
     `(num_patches, num_channels * patch_size * patch_size)`
     """
 
     image_grid_thw: torch.Tensor
     """Shape: `(num_images, 3)`
-    
+
     This should be in `(grid_t, grid_h, grid_w)` format.
     """
 
 
 class Qwen2VLVideoInputs(TypedDict):
     pixel_values_videos: torch.Tensor
-    """Shape: 
-    `(num_patches, 
+    """Shape:
+    `(num_patches,
       num_channels * temporal_patch_size * patch_size * patch_size)`
     """
 
     video_grid_thw: torch.Tensor
     """Shape: `(num_videos, 3)`
-    
+
     This should be in `(grid_t, grid_h, grid_w)` format.
     """
 

python/sglang/srt/server.py

@@ -759,7 +759,7 @@ class Engine:
 
         # before python program terminates, call shutdown implicitly. Therefore, users don't have to explicitly call .shutdown()
         atexit.register(self.shutdown)
-        
+
         # runtime server default log level is log
         # offline engine works in scripts, so we set it to error
 

python/sglang/test/simple_eval_common.py

@@ -320,7 +320,7 @@ jinja_env = jinja2.Environment(
 _message_template = """
 <div class="message {{ role }}">
     <div class="role">
-    {{ role }} 
+    {{ role }}
     {% if variant %}<span class="variant">({{ variant }})</span>{% endif %}
     </div>
     <div class="content">

python/sglang/test/simple_eval_humaneval.py

@@ -2,8 +2,8 @@
 
 """
 HumanEval: Evaluating Large Language Models Trained on Code
-Mark Chen and Jerry Tworek and Heewoo Jun and Qiming Yuan and Henrique Ponde de Oliveira Pinto and Jared Kaplan and Harri Edwards and Yuri Burda and Nicholas Joseph and Greg Brockman and Alex Ray and Raul Puri and Gretchen Krueger and Michael Petrov and Heidy Khlaaf and Girish Sastry and Pamela Mishkin and Brooke Chan and Scott Gray and Nick Ryder and Mikhail Pavlov and Alethea Power and Lukasz Kaiser and Mohammad Bavarian and Clemens Winter and Philippe Tillet and Felipe Petroski Such and Dave Cummings and Matthias Plappert and Fotios Chantzis and Elizabeth Barnes and Ariel Herbert-Voss and William Hebgen Guss and Alex Nichol and Alex Paino and Nikolas Tezak and Jie Tang and Igor Babuschkin and Suchir Balaji and Shantanu Jain and William Saunders and Christopher Hesse and Andrew N. Carr and Jan Leike and Josh Achiam and Vedant Misra and Evan Morikawa and Alec Radford and Matthew Knight and Miles Brundage and Mira Murati and Katie Mayer and Peter Welinder and Bob McGrew and Dario Amodei and Sam McCandlish and Ilya Sutskever and Wojciech Zaremba 
-https://arxiv.org/abs/2107.03374 https://github.com/openai/human-eval/ 
+Mark Chen and Jerry Tworek and Heewoo Jun and Qiming Yuan and Henrique Ponde de Oliveira Pinto and Jared Kaplan and Harri Edwards and Yuri Burda and Nicholas Joseph and Greg Brockman and Alex Ray and Raul Puri and Gretchen Krueger and Michael Petrov and Heidy Khlaaf and Girish Sastry and Pamela Mishkin and Brooke Chan and Scott Gray and Nick Ryder and Mikhail Pavlov and Alethea Power and Lukasz Kaiser and Mohammad Bavarian and Clemens Winter and Philippe Tillet and Felipe Petroski Such and Dave Cummings and Matthias Plappert and Fotios Chantzis and Elizabeth Barnes and Ariel Herbert-Voss and William Hebgen Guss and Alex Nichol and Alex Paino and Nikolas Tezak and Jie Tang and Igor Babuschkin and Suchir Balaji and Shantanu Jain and William Saunders and Christopher Hesse and Andrew N. Carr and Jan Leike and Josh Achiam and Vedant Misra and Evan Morikawa and Alec Radford and Matthew Knight and Miles Brundage and Mira Murati and Katie Mayer and Peter Welinder and Bob McGrew and Dario Amodei and Sam McCandlish and Ilya Sutskever and Wojciech Zaremba
+https://arxiv.org/abs/2107.03374 https://github.com/openai/human-eval/
 """
 
 import random

python/sglang/test/simple_eval_mgsm.py

@@ -1,10 +1,10 @@
 # Adapted from https://github.com/openai/simple-evals/
 
 """
-MGSM: Multilingual Grade School Math Benchmark (MGSM) is a benchmark of grade-school math problems. 
+MGSM: Multilingual Grade School Math Benchmark (MGSM) is a benchmark of grade-school math problems.
 Language Models are Multilingual Chain-of-Thought Reasoners
 Freda Shi, Mirac Suzgun, Markus Freitag, Xuezhi Wang, Suraj Srivats, Soroush Vosoughi, Hyung Won Chung, Yi Tay, Sebastian Ruder, Denny Zhou, Dipanjan Das, Jason Wei
-https://arxiv.org/abs/2210.03057 reference: https://github.com/google-research/url-nlp 
+https://arxiv.org/abs/2210.03057 reference: https://github.com/google-research/url-nlp
 """
 
 import re