Upload qwen2_tool_parser.py
Browse files- qwen2_tool_parser.py +122 -0
qwen2_tool_parser.py
ADDED
@@ -0,0 +1,122 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import json
|
2 |
+
import re
|
3 |
+
from typing import Dict, List, Sequence, Union
|
4 |
+
|
5 |
+
import partial_json_parser
|
6 |
+
from partial_json_parser.core.options import Allow
|
7 |
+
|
8 |
+
from vllm.entrypoints.openai.protocol import (ChatCompletionRequest,
|
9 |
+
DeltaFunctionCall, DeltaMessage,
|
10 |
+
DeltaToolCall,
|
11 |
+
ExtractedToolCallInformation,
|
12 |
+
FunctionCall, ToolCall)
|
13 |
+
from vllm.entrypoints.openai.tool_parsers.abstract_tool_parser import (
|
14 |
+
ToolParser, ToolParserManager)
|
15 |
+
from vllm.logger import init_logger
|
16 |
+
from vllm.transformers_utils.tokenizer import AnyTokenizer, MistralTokenizer
|
17 |
+
from vllm.utils import random_uuid
|
18 |
+
|
19 |
+
logger = init_logger(__name__)
|
20 |
+
|
21 |
+
|
22 |
+
@ToolParserManager.register_module("qwen2")
|
23 |
+
class Qwen2ToolParser(ToolParser):
|
24 |
+
|
25 |
+
def __init__(self, tokenizer: AnyTokenizer):
|
26 |
+
super().__init__(tokenizer)
|
27 |
+
|
28 |
+
if isinstance(self.model_tokenizer, MistralTokenizer):
|
29 |
+
logger.error(
|
30 |
+
"Detected Mistral tokenizer when using a Qwen2.5 model")
|
31 |
+
self.model_tokenizer = self.model_tokenizer.tokenizer
|
32 |
+
|
33 |
+
self.current_tool_name_sent: bool = False
|
34 |
+
self.prev_tool_call_arr: List[Dict] = []
|
35 |
+
self.current_tool_id: int = -1
|
36 |
+
self.streamed_args_for_tool: List[str] = [
|
37 |
+
] # map what has been streamed for each tool so far to a list
|
38 |
+
|
39 |
+
self.tool_call_start_token: str = "<tool_call>"
|
40 |
+
self.tool_call_end_token: str = "</tool_call>"
|
41 |
+
|
42 |
+
self.tool_call_regex = re.compile(
|
43 |
+
r"<tool_call>(.*?)</tool_call>", re.DOTALL)
|
44 |
+
self.scratch_pad_regex = re.compile(
|
45 |
+
r"<scratch_pad>(.*?)</scratch_pad>", re.DOTALL)
|
46 |
+
|
47 |
+
if not self.model_tokenizer:
|
48 |
+
raise ValueError(
|
49 |
+
"The model tokenizer must be passed to the ToolParser "
|
50 |
+
"constructor during construction.")
|
51 |
+
self.tool_call_start_token_id = self.vocab.get(
|
52 |
+
self.tool_call_start_token)
|
53 |
+
self.tool_call_end_token_id = self.vocab.get(self.tool_call_end_token)
|
54 |
+
if (self.tool_call_start_token_id is None
|
55 |
+
or self.tool_call_end_token_id is None):
|
56 |
+
raise RuntimeError(
|
57 |
+
"Qwen2.5 Tool parser could not locate tool call start/end "
|
58 |
+
"tokens in the tokenizer!")
|
59 |
+
|
60 |
+
def extract_tool_calls(
|
61 |
+
self,
|
62 |
+
model_output: str,
|
63 |
+
request: ChatCompletionRequest,
|
64 |
+
) -> ExtractedToolCallInformation:
|
65 |
+
|
66 |
+
# sanity check; avoid unnecessary processing
|
67 |
+
if self.tool_call_start_token not in model_output:
|
68 |
+
return ExtractedToolCallInformation(tools_called=False,
|
69 |
+
tool_calls=[],
|
70 |
+
content=model_output)
|
71 |
+
|
72 |
+
else:
|
73 |
+
|
74 |
+
try:
|
75 |
+
# find all tool calls between "<tool_call>" and "</tool_call>"
|
76 |
+
# the other is None
|
77 |
+
function_call_strs = (
|
78 |
+
self.tool_call_regex.findall(model_output))
|
79 |
+
|
80 |
+
# load the JSON, and then use it to build the Function and
|
81 |
+
# Tool Call
|
82 |
+
raw_function_calls = json.loads(function_call_strs[0])
|
83 |
+
|
84 |
+
tool_calls = [
|
85 |
+
ToolCall(
|
86 |
+
type="function",
|
87 |
+
function=FunctionCall(
|
88 |
+
name=function_call["tool_name"],
|
89 |
+
# function call args are JSON but as a string
|
90 |
+
arguments=json.dumps(function_call["parameters"], ensure_ascii=False)
|
91 |
+
)
|
92 |
+
)
|
93 |
+
for function_call in raw_function_calls
|
94 |
+
]
|
95 |
+
|
96 |
+
content = model_output[:model_output.
|
97 |
+
find(self.tool_call_start_token)]
|
98 |
+
return ExtractedToolCallInformation(
|
99 |
+
tools_called=True,
|
100 |
+
tool_calls=tool_calls,
|
101 |
+
content=content if content else None)
|
102 |
+
|
103 |
+
except Exception:
|
104 |
+
logger.exception(
|
105 |
+
"Error in extracting tool call from response.")
|
106 |
+
return ExtractedToolCallInformation(tools_called=False,
|
107 |
+
tool_calls=[],
|
108 |
+
content=model_output)
|
109 |
+
|
110 |
+
# for streamed parsing
|
111 |
+
def extract_tool_calls_streaming(
|
112 |
+
self,
|
113 |
+
previous_text: str,
|
114 |
+
current_text: str,
|
115 |
+
delta_text: str,
|
116 |
+
previous_token_ids: Sequence[int],
|
117 |
+
current_token_ids: Sequence[int],
|
118 |
+
delta_token_ids: Sequence[int],
|
119 |
+
request: ChatCompletionRequest,
|
120 |
+
) -> Union[DeltaMessage, None]:
|
121 |
+
|
122 |
+
pass
|