@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / f694d43b33
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

[Bugfix][test] Use Salesforce/wikitext for ppl tests (#45913)

Co-authored-by: wentian-byte <192079369+wentian-byte@users.noreply.github.com>

wentian-byte committed Jun 17, 2026 at 18:37 UTC f694d43b33503f18dddd68857688c6826890a847
1 file changed +1 -1
tests/models/language/generation_ppl_test/ppl_utils.py
+1 -1
@@ -30,7 +30,7 @@ def wikitext_ppl_test(
30 ):
31 vllm_extra_kwargs = get_vllm_extra_kwargs(model_info, vllm_extra_kwargs)
32
33 - dataset = load_dataset("wikitext", "wikitext-2-raw-v1", split="test")
33 + dataset = load_dataset("Salesforce/wikitext", "wikitext-2-raw-v1", split="test")
34
35 with vllm_runner(
36 model_info.name,