tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

data_parallel_size in in VllmserveCliArgs (#3074)

Browse Source 
* data_parallel_size in in VllmserveCliArgs

* moved to 43
This commit is contained in:
VED
2025-08-18 18:14:37 +05:30
committed by GitHub
parent 0eef385b1a
commit c10eb811fa
1 changed files with 6 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/axolotl/cli/args.py
View File

@@ -40,6 +40,12 @@ class VllmServeCliArgs:
default=None, default=None,
metadata={"help": "Number of tensor parallel workers to use."}, metadata={"help": "Number of tensor parallel workers to use."},
) )
data_parallel_size: Optional[int] = field(
default=None,
metadata={
"help": "Number of data parallel workers to use for vLLM serving. This controls how many model replicas are used for parallel inference."
},
)
host: Optional[str] = field( host: Optional[str] = field(
default=None, # nosec B104 default=None, # nosec B104
metadata={"help": "Host address to run the server on."}, metadata={"help": "Host address to run the server on."},