Fix the chat template for llava-v1.6-34b & format code (#177)

test/lang/test_srt_backend.py

@@ -1,6 +1,7 @@
 """
 python3 -m sglang.launch_server --model-path meta-llama/Llama-2-7b-chat-hf --port 30000
 """
+
 import json
 import unittest
 

test/srt/model/bench_llama_low_api.py

@@ -66,9 +66,9 @@ class BenchBatch:
             p_idx = prefix_req_idx[i // fork_num].item()
             n_idx = self.req_pool_indices[i].item()
             req_to_token[n_idx, :prefix_len] = req_to_token[p_idx, :prefix_len]
-            req_to_token[
-                n_idx, prefix_len : prefix_len + extend_len
-            ] = self.out_cache_loc[i * extend_len : (i + 1) * extend_len]
+            req_to_token[n_idx, prefix_len : prefix_len + extend_len] = (
+                self.out_cache_loc[i * extend_len : (i + 1) * extend_len]
+            )
 
     def update_decode(self, predict_ids, batch_size):
         assert predict_ids.shape[0] == batch_size
@@ -81,9 +81,9 @@ class BenchBatch:
             self.out_cache_cont_start,
             self.out_cache_cont_end,
         ) = self.token_to_kv_pool.alloc_contiguous(batch_size)
-        self.req_to_token_pool.req_to_token[
-            self.req_pool_indices, self.seq_lens
-        ] = self.out_cache_loc
+        self.req_to_token_pool.req_to_token[self.req_pool_indices, self.seq_lens] = (
+            self.out_cache_loc
+        )
         self.seq_lens.add_(1)
 
 

test/srt/test_jump_forward.py

@@ -7,6 +7,7 @@ from sglang.test.test_utils import (
     add_common_sglang_args_and_parse,
     select_sglang_backend,
 )
+
 import sglang as sgl
 
 IP_REGEX = r"((25[0-5]|2[0-4]\d|[01]?\d\d?)\.){3}(25[0-5]|2[0-4]\d|[01]?\d\d?)"

test/srt/test_openai_server.py

@@ -155,7 +155,8 @@ def test_chat_completion_stream(args):
 def test_regex(args):
     client = openai.Client(api_key="EMPTY", base_url=args.base_url)
 
-    regex = (r"""\{\n"""
+    regex = (
+        r"""\{\n"""
         + r"""   "name": "[\w]+",\n"""
         + r"""   "population": "[\w\d\s]+"\n"""
         + r"""\}"""