Review changes

Signed-off-by: Xin Yang <[email protected]>
vllm-project · Dec 3, 2024 · e4b30d0 · e4b30d0
1 parent 3b88608
commit e4b30d0
Show file tree

Hide file tree

Showing 5 changed files with 2 additions and 15 deletions.
diff --git a/tests/entrypoints/test_chat_utils.py b/tests/entrypoints/test_chat_utils.py
@@ -46,7 +46,6 @@ def phi3v_tokenizer():
         tokenizer_id=PHI3V_MODEL_ID,
         enable_lora=False,
         max_num_seqs=5,
-        max_loras=0,
         max_input_length=None,
     )
 
@@ -71,7 +70,6 @@ def mllama_tokenizer():
         MLLAMA_MODEL_ID,
         enable_lora=False,
         max_num_seqs=5,
-        max_loras=0,
         max_input_length=None,
     )
 
@@ -684,7 +682,6 @@ def get_conversation(is_hf: bool):
         MLLAMA_MODEL_ID,
         enable_lora=False,
         max_num_seqs=5,
-        max_loras=0,
         max_input_length=None,
     )
     tokenizer = tokenizer_group.tokenizer
@@ -731,7 +728,6 @@ def test_resolve_content_format_hf_defined(model, expected_format):
         model,
         enable_lora=False,
         max_num_seqs=5,
-        max_loras=0,
         max_input_length=None,
     )
     tokenizer = tokenizer_group.tokenizer
@@ -781,7 +777,6 @@ def test_resolve_content_format_examples(template_path, expected_format):
         PHI3V_MODEL_ID,
         enable_lora=False,
         max_num_seqs=5,
-        max_loras=0,
         max_input_length=None,
     )
     dummy_tokenizer = tokenizer_group.tokenizer

diff --git a/tests/test_cache_block_hashing.py b/tests/test_cache_block_hashing.py
@@ -49,7 +49,6 @@ def test_auto_prefix_caching(model: str, block_size: int, max_num_seqs: int,
         tokenizer_id="facebook/opt-125m",
         enable_lora=False,
         max_num_seqs=max_num_seqs,
-        max_loras=0,
         max_input_length=None,
     )
 

diff --git a/tests/tokenization/test_tokenizer_group.py b/tests/tokenization/test_tokenizer_group.py
@@ -36,7 +36,6 @@ async def test_tokenizer_group(tokenizer_group_type):
         tokenizer_id="gpt2",
         enable_lora=False,
         max_num_seqs=1,
-        max_loras=0,
         max_input_length=None,
     )
     assert reference_tokenizer.encode("prompt") == tokenizer_group.encode(
@@ -61,7 +60,6 @@ async def test_tokenizer_group_pool(tokenizer_group_type):
         tokenizer_id="gpt2",
         enable_lora=False,
         max_num_seqs=1,
-        max_loras=0,
         max_input_length=None,
     )
     # Send multiple requests to the tokenizer group pool
@@ -104,7 +102,6 @@ class EnvVarCheckerRayTokenizerGroupPool(RayTokenizerGroupPool):
         tokenizer_id="gpt2",
         enable_lora=False,
         max_num_seqs=1,
-        max_loras=0,
         max_input_length=None)
     with pytest.raises(AssertionError):
         tokenizer_pool.ping()
@@ -116,7 +113,6 @@ class EnvVarCheckerRayTokenizerGroupPool(RayTokenizerGroupPool):
             tokenizer_id="gpt2",
             enable_lora=False,
             max_num_seqs=1,
-            max_loras=0,
             max_input_length=None)
         tokenizer_pool.ping()
 
@@ -154,7 +150,6 @@ class FailingRayTokenizerGroupPool(RayTokenizerGroupPool):
         tokenizer_id="gpt2",
         enable_lora=False,
         max_num_seqs=1,
-        max_loras=0,
         max_input_length=None,
         fail_at=fail_at)
     tokenizer_actors = tokenizer_group_pool.tokenizer_actors.copy()
@@ -182,7 +177,6 @@ class FailingRayTokenizerGroupPool(RayTokenizerGroupPool):
         tokenizer_id="gpt2",
         enable_lora=False,
         max_num_seqs=1,
-        max_loras=0,
         max_input_length=None,
         fail_at=fail_at)
 
@@ -204,7 +198,6 @@ class FailingRayTokenizerGroupPool(RayTokenizerGroupPool):
         tokenizer_id="gpt2",
         enable_lora=False,
         max_num_seqs=1,
-        max_loras=0,
         max_input_length=2,
         fail_at=fail_at)
     tokenizer_actors = tokenizer_group_pool.tokenizer_actors.copy()

diff --git a/vllm/transformers_utils/tokenizer_group/ray_tokenizer_group.py b/vllm/transformers_utils/tokenizer_group/ray_tokenizer_group.py
@@ -51,7 +51,7 @@ def from_config(cls, tokenizer_pool_config: Optional[TokenizerPoolConfig],
         return cls(**init_kwargs)
 
     def __init__(self, tokenizer_id: str, enable_lora: bool, max_num_seqs: int,
-                 max_loras: int, max_input_length: Optional[int],
+                 max_loras: Optional[int], max_input_length: Optional[int],
                  num_actors: int, ray_actor_options: dict, **tokenizer_config):
         # Store a local copy of the TokenizerGroup for quick access
         # to underlying HF tokenizers.

diff --git a/vllm/transformers_utils/tokenizer_group/tokenizer_group.py b/vllm/transformers_utils/tokenizer_group/tokenizer_group.py
@@ -15,7 +15,7 @@ class TokenizerGroup(BaseTokenizerGroup):
     """A group of tokenizers that can be used for LoRA adapters."""
 
     def __init__(self, tokenizer_id: str, enable_lora: bool, max_num_seqs: int,
-                 max_loras: int, max_input_length: Optional[int],
+                 max_loras: Optional[int], max_input_length: Optional[int],
                  **tokenizer_config):
         self.tokenizer_id = tokenizer_id
         self.tokenizer_config = tokenizer_config