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, 22:00 UTC
[ci][build] fix commit id (#6420)
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
youkaichao committed
Jul 14, 2024 at 07:16 UTC
ccd3c045710323102ebd9eab6d2e192c0fd6e509
3 files changed
+13
-3
.buildkite/test-pipeline.yaml
+3
-1
@@ -194,7 +194,9 @@ steps:
194
command: pytest -v -s test_logits_processor.py
195
196
- label: Utils Test
197
- command: pytest -v -s test_utils.py
197
+ commands:
198
+ - pytest -v -s test_utils.py
199
+ - pytest -v -s test_embedded_commit.py
200
201
- label: Worker Test
202
mirror_hardwares: [amd]
Dockerfile
+3
@@ -88,6 +88,9 @@ ENV NVCC_THREADS=$nvcc_threads
88
# make sure punica kernels are built (for LoRA)
89
ENV VLLM_INSTALL_PUNICA_KERNELS=1
90
91
+ARG buildkite_commit
92
+ENV BUILDKITE_COMMIT=${buildkite_commit}
93
+
94
ARG USE_SCCACHE
95
# if USE_SCCACHE is set, use sccache to speed up compilation
96
RUN --mount=type=cache,target=/root/.cache/pip \
setup.py
+7
-2
@@ -30,8 +30,13 @@ logger = logging.getLogger(__name__)
30
31
def embed_commit_hash():
32
try:
33
- commit_id = subprocess.check_output(["git", "rev-parse", "HEAD"],
34
- encoding="utf-8").strip()
33
+ if "BUILDKITE_COMMIT" in os.environ:
34
+ # ci build
35
+ commit_id = os.environ["BUILDKITE_COMMIT"]
36
+ else:
37
+ commit_id = subprocess.check_output(["git", "rev-parse", "HEAD"],
38
+ encoding="utf-8").strip()
39
+
40
commit_contents = f'__commit__ = "{commit_id}"\n'
41
42
version_file = os.path.join(ROOT_DIR, "vllm", "commit_id.py")