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 13, 20:45 UTC
[CI/Build] Fix OpenAI API correctness on AMD CI (#28022)
Signed-off-by: zhewenli <zhewenli@meta.com>
Zhewen Li committed
Nov 3, 2025 at 23:20 UTC
53f6e81dfd9cdba797ddade119a5e33389a35957
1 file changed
+4
-3
.buildkite/test-amd.yaml
+4
-3
@@ -629,15 +629,16 @@ steps:
629
630
- label: OpenAI API correctness # 22min
631
timeout_in_minutes: 30
632
- mirror_hardwares: [amdexperimental]
632
+ mirror_hardwares: [amdexperimental, amdproduction]
633
agent_pool: mi325_1
634
# grade: Blocking
635
source_file_dependencies:
636
- csrc/
637
- vllm/entrypoints/openai/
638
- vllm/model_executor/models/whisper.py
639
- commands: # LMEval+Transcription WER check
640
- - pytest -s entrypoints/openai/correctness/
639
+ commands: # LMEval
640
+ # Transcription WER check is skipped because encoder-decoder models are not supported on ROCm, see https://github.com/vllm-project/vllm/issues/27442
641
+ - pytest -s entrypoints/openai/correctness/ --ignore entrypoints/openai/correctness/test_transcription_api_correctness.py
642
643
- label: OpenAI-Compatible Tool Use # 23 min
644
timeout_in_minutes: 35