@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 7187507301
Fork 0

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 16, 23:15 UTC

[ci][test] fix ca test in main (#5746)

youkaichao committed Jun 21, 2024 at 14:04 UTC 7187507301aa8361407e04be42d0d50680891493
1 file changed +3
.buildkite/test-pipeline.yaml
+3
@@ -197,6 +197,9 @@ steps:
197 gpu: a100
198 num_gpus: 4
199 commands:
200 + # FIXIT: find out which code initialize cuda before running the test
201 + # before the fix, we need to use spawn to test it
202 + - export VLLM_WORKER_MULTIPROC_METHOD=spawn
203 # NOTE: don't test llama model here, it seems hf implementation is buggy
204 # see https://github.com/vllm-project/vllm/pull/5689 for details
205 - pytest -v -s distributed/test_custom_all_reduce.py