biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] Refactor tokenizer interface (#29693)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-11-29 20:02:21 +08:00
committed by GitHub
parent f223ed4181
commit 34a984274e
119 changed files with 752 additions and 821 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/tool_use/test_ernie45_moe_tool_parser.py
View File

@@ -14,8 +14,9 @@ from vllm.entrypoints.openai.protocol import (
ToolCall,
)
from vllm.entrypoints.openai.tool_parsers.ernie45_tool_parser import Ernie45ToolParser
from vllm.tokenizers import TokenizerLike
from vllm.transformers_utils.detokenizer_utils import detokenize_incrementally
from vllm.transformers_utils.tokenizer import AnyTokenizer, get_tokenizer
from vllm.transformers_utils.tokenizer import get_tokenizer
# Use a common model that is likely to be available
MODEL = "baidu/ERNIE-4.5-21B-A3B-Thinking"
@@ -173,7 +174,7 @@ def test_extract_tool_calls(
def stream_delta_message_generator(
ernie45_tool_parser: Ernie45ToolParser,
ernie45_tokenizer: AnyTokenizer,
ernie45_tokenizer: TokenizerLike,
model_output: str,
request: ChatCompletionRequest | None = None,
) -> Generator[DeltaMessage, None, None]:

7
tests/tool_use/test_jamba_tool_parser.py
View File

@@ -10,8 +10,9 @@ from partial_json_parser.core.options import Allow
from vllm.entrypoints.openai.protocol import DeltaMessage, FunctionCall, ToolCall
from vllm.entrypoints.openai.tool_parsers.jamba_tool_parser import JambaToolParser
from vllm.tokenizers import TokenizerLike
from vllm.transformers_utils.detokenizer_utils import detokenize_incrementally
from vllm.transformers_utils.tokenizer import AnyTokenizer, get_tokenizer
from vllm.transformers_utils.tokenizer import get_tokenizer
pytestmark = pytest.mark.cpu_test
@@ -44,7 +45,9 @@ def assert_tool_calls(
def stream_delta_message_generator(
jamba_tool_parser: JambaToolParser, jamba_tokenizer: AnyTokenizer, model_output: str
jamba_tool_parser: JambaToolParser,
jamba_tokenizer: TokenizerLike,
model_output: str,
) -> Generator[DeltaMessage, None, None]:
all_token_ids = jamba_tokenizer.encode(model_output, add_special_tokens=False)

5
tests/tool_use/test_qwen3coder_tool_parser.py
View File

@@ -17,8 +17,9 @@ from vllm.entrypoints.openai.tool_parsers.qwen3coder_tool_parser import (
Qwen3CoderToolParser,
)
from vllm.entrypoints.openai.tool_parsers.qwen3xml_tool_parser import Qwen3XMLToolParser
from vllm.tokenizers import TokenizerLike
from vllm.transformers_utils.detokenizer_utils import detokenize_incrementally
from vllm.transformers_utils.tokenizer import AnyTokenizer, get_tokenizer
from vllm.transformers_utils.tokenizer import get_tokenizer
pytestmark = pytest.mark.cpu_test
@@ -104,7 +105,7 @@ def assert_tool_calls(
def stream_delta_message_generator(
qwen3_tool_parser,
qwen3_tokenizer: AnyTokenizer,
qwen3_tokenizer: TokenizerLike,
model_output: str,
request: ChatCompletionRequest | None = None,
) -> Generator[DeltaMessage, None, None]:

5
tests/tool_use/test_seed_oss_tool_parser.py
View File

@@ -15,8 +15,9 @@ from vllm.entrypoints.openai.protocol import (
ToolCall,
)
from vllm.entrypoints.openai.tool_parsers.seed_oss_tool_parser import SeedOssToolParser
from vllm.tokenizers import TokenizerLike
from vllm.transformers_utils.detokenizer_utils import detokenize_incrementally
from vllm.transformers_utils.tokenizer import AnyTokenizer, get_tokenizer
from vllm.transformers_utils.tokenizer import get_tokenizer
pytestmark = pytest.mark.cpu_test
@@ -256,7 +257,7 @@ def test_streaming_tool_calls_no_tools(seed_oss_tool_parser):
def stream_delta_message_generator(
seed_oss_tool_parser: SeedOssToolParser,
seed_oss_tokenizer: AnyTokenizer,
seed_oss_tokenizer: TokenizerLike,
model_output: str,
request: ChatCompletionRequest | None = None,
) -> Generator[DeltaMessage, None, None]:

5
tests/tool_use/test_xlam_tool_parser.py
View File

@@ -13,8 +13,9 @@ from vllm.entrypoints.openai.protocol import (
ToolCall,
)
from vllm.entrypoints.openai.tool_parsers.xlam_tool_parser import xLAMToolParser
from vllm.tokenizers import TokenizerLike
from vllm.transformers_utils.detokenizer_utils import detokenize_incrementally
from vllm.transformers_utils.tokenizer import AnyTokenizer, get_tokenizer
from vllm.transformers_utils.tokenizer import get_tokenizer
pytestmark = pytest.mark.cpu_test
@@ -49,7 +50,7 @@ def assert_tool_calls(
def stream_delta_message_generator(
xlam_tool_parser: xLAMToolParser,
xlam_tokenizer: AnyTokenizer,
xlam_tokenizer: TokenizerLike,
model_output: str,
request: ChatCompletionRequest | None = None,
) -> Generator[DeltaMessage, None, None]: