A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Spec Decode] Fix offline spec_decode.py (#24257)
Signed-off-by: Ekagra Ranjan <3116519+ekagra-ranjan@users.noreply.github.com> Co-authored-by: Roger Wang <hey@rogerw.io>
Ekagra Ranjan committed
Sep 8, 2025 at 16:44 UTC
41183c1fe09c60cd77d683e64895c08f0d84b693
1 file changed
+4
vllm/benchmarks/datasets.py
+4
@@ -1310,6 +1310,10 @@ def add_dataset_parser(parser: FlexibleArgumentParser):
1310
1311
1312
def get_samples(args, tokenizer) -> list[SampleRequest]:
1313
+
1314
+ if not hasattr(args, "request_id_prefix"):
1315
+ args.request_id_prefix = ""
1316
+
1317
if args.dataset_name == "custom":
1318
dataset = CustomDataset(dataset_path=args.dataset_path)
1319
input_requests = dataset.sample(