You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: components/backends/vllm/src/dynamo/vllm/args.py
+9-1Lines changed: 9 additions & 1 deletion
Original file line number
Diff line number
Diff line change
@@ -47,6 +47,7 @@ class Config:
47
47
migration_limit: int=0
48
48
kv_port: Optional[int] =None
49
49
port_range: DynamoPortRange
50
+
custom_jinja_template: Optional[str] =None
50
51
51
52
# mirror vLLM
52
53
model: str
@@ -100,7 +101,7 @@ def parse_args() -> Config:
100
101
help="List of connectors to use in order (e.g., --connector nixl lmcache). "
101
102
"Options: nixl, lmcache, kvbm, null, none. Default: nixl. Order will be preserved in MultiConnector.",
102
103
)
103
-
# To avoid name conflicts with different backends, adoped prefix "dyn-" for dynamo specific args
104
+
# To avoid name conflicts with different backends, adopted prefix "dyn-" for dynamo specific args
104
105
parser.add_argument(
105
106
"--dyn-tool-call-parser",
106
107
type=str,
@@ -115,6 +116,12 @@ def parse_args() -> Config:
115
116
choices=get_reasoning_parser_names(),
116
117
help="Reasoning parser name for the model.",
117
118
)
119
+
parser.add_argument(
120
+
"--custom-jinja-template",
121
+
type=str,
122
+
default=None,
123
+
help="Path to a custom Jinja template file to override the model's default chat template. This template will take precedence over any template found in the model repository.",
0 commit comments