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 4, 15:00 UTC
[ROCm][CI] Fix `trust_remote_code` AttributeError in EAGLE3 acceptance length test (#40306)
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
Andreas Karatzas committed
Apr 24, 2026 at 21:59 UTC
428b988c98a0dee06c47d4a70858317b60169461
1 file changed
+3
-2
tests/v1/spec_decode/test_acceptance_length.py
+3
-2
@@ -165,6 +165,7 @@ def get_mt_bench_prompts(
165
no_stream=True,
166
disable_shuffle=False,
167
skip_chat_template=False,
168
+ trust_remote_code=False,
169
)
170
samples = get_samples(args, tokenizer)
171
prompt_ids = [
@@ -210,8 +211,8 @@ def extract_acceptance_metrics(metrics, num_spec_tokens: int) -> dict:
211
212
@large_gpu_mark(min_gb=40)
213
@pytest.mark.skipif(
213
- not current_platform.is_cuda(),
214
- reason="This test is only supported on CUDA platform.",
214
+ not current_platform.is_cuda_alike(),
215
+ reason="This test is only supported on CUDA-alike platforms.",
216
)
217
@pytest.mark.parametrize(
218
"model_config",