@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 4fadf9c92c
Fork 0

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

[ROCm][CI] fix multimodel run cmds (#45858)

Signed-off-by: Divakar Verma <divakar.verma@amd.com>

Divakar Verma committed Jun 16, 2026 at 16:31 UTC 4fadf9c92c203ac4745caadfaac0ff4e6fe0806e
1 file changed +5 -7
.buildkite/test-amd.yaml
+5 -7
@@ -1783,10 +1783,9 @@ steps:
1783 - tests/models/multimodal/generation
1784 - tests/models/multimodal/test_mapping.py
1785 commands:
1786 - - uv pip install --system --no-build-isolation 'git+https://github.com/AndreasKaratzas/mamba@rocm-7.0-v2.3.0'
1787 - - uv pip install --system --no-build-isolation 'git+https://github.com/Dao-AILab/causal-conv1d@v1.6.0'
1788 - - pytest -v -s models/language/generation -m hybrid_model --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
1789 -
1786 + - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
1787 + - pytest -v -s models/multimodal/generation -m 'not core_model' --ignore models/multimodal/generation/test_common.py
1788 + - pytest -v -s models/multimodal/test_mapping.py
1789
1790 - label: Multi-Modal Models (Extended Generation 2) # TBD
1791 timeout_in_minutes: 180
@@ -1798,9 +1797,8 @@ steps:
1797 - vllm/
1798 - tests/models/multimodal/generation
1799 commands:
1801 - - uv pip install --system --no-build-isolation 'git+https://github.com/AndreasKaratzas/mamba@rocm-7.0-v2.3.0'
1802 - - uv pip install --system --no-build-isolation 'git+https://github.com/Dao-AILab/causal-conv1d@v1.6.0'
1803 - - pytest -v -s models/language/generation -m '(not core_model) and (not hybrid_model)'
1800 + - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
1801 + - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=0) and not core_model'
1802
1803
1804 - label: Multi-Modal Models (Extended Generation 3) # TBD