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, 05:45 UTC
Build tpu image in release pipeline (#10936)
Signed-off-by: Richard Liu <ricliu@google.com> Co-authored-by: Kevin H. Luu <kevin@anyscale.com>
Richard Liu committed
Dec 9, 2024 at 15:07 UTC
5ed5d5f128d26a48c1b1db16c319fcb96c93799d
1 file changed
+16
.buildkite/release-pipeline.yaml
+16
@@ -39,3 +39,19 @@ steps:
39
- "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
40
- "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.1.0 --tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT --target vllm-openai --progress plain ."
41
- "docker push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT"
42
+
43
+ - label: "Build and publish TPU release image"
44
+ depends_on: ~
45
+ if: build.env("NIGHTLY") == "1"
46
+ agents:
47
+ queue: tpu_queue_postmerge
48
+ commands:
49
+ - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --tag vllm/vllm-tpu:nightly --tag vllm/vllm-tpu:$BUILDKITE_COMMIT --progress plain -f Dockerfile.tpu ."
50
+ - "docker push vllm/vllm-tpu:nightly"
51
+ - "docker push vllm/vllm-tpu:$BUILDKITE_COMMIT"
52
+ plugins:
53
+ - docker-login#v3.0.0:
54
+ username: vllm
55
+ password-env: DOCKERHUB_TOKEN
56
+ env:
57
+ DOCKER_BUILDKIT: "1"