[Renderer] Move InputPreprocessor into Renderer (1/2) (#34510)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
This commit is contained in:
Cyrus Leung
2026-02-14 10:14:21 -08:00
committed by GitHub
parent b3c14229b0
commit 73391a1baa
39 changed files with 456 additions and 458 deletions
+2 -1
View File
@@ -54,6 +54,7 @@ class MockModelConfig:
media_io_kwargs: dict[str, dict[str, Any]] = field(default_factory=dict)
skip_tokenizer_init = False
is_encoder_decoder: bool = False
is_multimodal_model: bool = False
def get_diff_sampling_param(self):
return self.diff_sampling_param or {}
@@ -67,7 +68,7 @@ class MockVllmConfig:
def _build_renderer(model_config: MockModelConfig):
_, tokenizer_name, _, kwargs = tokenizer_args_from_config(model_config)
return HfRenderer(
return HfRenderer.from_config(
MockVllmConfig(model_config),
tokenizer_kwargs={**kwargs, "tokenizer_name": tokenizer_name},
)
@@ -53,6 +53,7 @@ class MockModelConfig:
media_io_kwargs: dict[str, dict[str, Any]] = field(default_factory=dict)
skip_tokenizer_init = False
is_encoder_decoder: bool = False
is_multimodal_model: bool = False
def get_diff_sampling_param(self):
return self.diff_sampling_param or {}
@@ -78,7 +79,7 @@ def _build_serving_completion(engine: AsyncLLM) -> OpenAIServingCompletion:
def _build_renderer(model_config: MockModelConfig):
_, tokenizer_name, _, kwargs = tokenizer_args_from_config(model_config)
return HfRenderer(
return HfRenderer.from_config(
MockVllmConfig(model_config),
tokenizer_kwargs={**kwargs, "tokenizer_name": tokenizer_name},
)
@@ -52,6 +52,7 @@ class MockModelConfig:
generation_config: str = "auto"
skip_tokenizer_init: bool = False
is_encoder_decoder: bool = False
is_multimodal_model: bool = False
def get_diff_sampling_param(self):
return self.diff_sampling_param or {}
@@ -95,7 +96,7 @@ def register_mock_resolver():
def _build_renderer(model_config: MockModelConfig):
_, tokenizer_name, _, kwargs = tokenizer_args_from_config(model_config)
return HfRenderer(
return HfRenderer.from_config(
MockVllmConfig(model_config),
tokenizer_kwargs={**kwargs, "tokenizer_name": tokenizer_name},
)
@@ -529,6 +529,7 @@ class MockModelConfig:
media_io_kwargs: dict[str, dict[str, Any]] = field(default_factory=dict)
skip_tokenizer_init: bool = False
is_encoder_decoder: bool = False
is_multimodal_model: bool = False
def get_diff_sampling_param(self):
return self.diff_sampling_param or {}
@@ -542,7 +543,7 @@ class MockVllmConfig:
def _build_renderer(model_config: MockModelConfig):
_, tokenizer_name, _, kwargs = tokenizer_args_from_config(model_config)
return HfRenderer(
return HfRenderer.from_config(
MockVllmConfig(model_config),
tokenizer_kwargs={**kwargs, "tokenizer_name": tokenizer_name},
)
@@ -756,9 +757,8 @@ async def test_serving_chat_mistral_token_ids_prompt_is_validated():
mock_tokenizer = MagicMock(spec=MistralTokenizer)
mock_renderer = MistralRenderer(
MockVllmConfig(mock_engine.model_config),
tokenizer_kwargs={},
tokenizer=mock_tokenizer,
)
mock_renderer._tokenizer = mock_tokenizer
# Force the Mistral chat template renderer to return token IDs.
# Choose a prompt length that is < max_model_len, but large enough that
# adding max_tokens should exceed the model context window.
@@ -798,9 +798,8 @@ async def test_serving_chat_mistral_token_ids_prompt_too_long_is_rejected():
mock_tokenizer = MagicMock(spec=MistralTokenizer)
mock_renderer = MistralRenderer(
MockVllmConfig(mock_engine.model_config),
tokenizer_kwargs={},
tokenizer=mock_tokenizer,
)
mock_renderer._tokenizer = mock_tokenizer
# prompt_token_ids length == max_model_len should be rejected for
# completion-like requests (ChatCompletionRequest).
mock_renderer.render_messages_async = AsyncMock(