minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[Fix][Eval] Keep --dataset-path scoped to longbench_v2 (#21156)

Browse Source
This commit is contained in:
Mohammad Miadh Angkad
2026-03-24 02:25:11 -07:00
committed by GitHub
parent 8d7453d8fb
commit 4fbb311234
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/test/run_eval.py
+2 -2
View File
@@ -118,7 +118,7 @@ def run_eval(args):
elif args.eval_name == "gpqa": elif args.eval_name == "gpqa":
from sglang.test.simple_eval_gpqa import GPQAEval from sglang.test.simple_eval_gpqa import GPQAEval
filename = getattr(args, "dataset_path", None) or ( filename = (
"https://openaipublic.blob.core.windows.net/simple-evals/gpqa_diamond.csv" "https://openaipublic.blob.core.windows.net/simple-evals/gpqa_diamond.csv"
) )
eval_obj = GPQAEval(filename, args.num_examples, args.num_threads) eval_obj = GPQAEval(filename, args.num_examples, args.num_threads)
@@ -130,7 +130,7 @@ def run_eval(args):
from sglang.test.simple_eval_longbench_v2 import LongBenchV2Eval from sglang.test.simple_eval_longbench_v2 import LongBenchV2Eval
# Default to HuggingFace dataset, can be overridden with --dataset-path # Default to HuggingFace dataset, can be overridden with --dataset-path
data_source = getattr(args, "dataset_path", None) data_source = args.dataset_path
categories = args.categories.split(",") if args.categories else None categories = args.categories.split(",") if args.categories else None
eval_obj = LongBenchV2Eval( eval_obj = LongBenchV2Eval(