Skip to content

Fix vllm batch docker image #463

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
yunfeng-scale merged 2 commits into from
Mar 7, 2024
Merged

Fix vllm batch docker image #463

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
fbe1e93
Fix vllm batch docker image
yunfeng-scale Mar 6, 2024
534c2d3
try again with 0.2.5
yunfeng-scale Mar 7, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 6 additions & 1 deletion model-engine/model_engine_server/inference/batch_inference/Dockerfile_vllm
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,10 +6,15 @@ RUN apt-get update && \
rm -rf /var/lib/apt/lists/* && \
apt-get clean

RUN pip uninstall torch -y
COPY model-engine/model_engine_server/inference/batch_inference/requirements.txt /workspace/requirements.txt
RUN pip install -r requirements.txt

RUN pip uninstall torch -y
RUN pip install torch==2.1.1 --index-url https://download.pytorch.org/whl/cu121

RUN pip uninstall xformers -y
RUN pip install xformers==0.0.23 --index-url https://download.pytorch.org/whl/cu121

RUN wget https://github.com/peak/s5cmd/releases/download/v2.2.1/s5cmd_2.2.1_Linux-64bit.tar.gz
RUN tar -xvzf s5cmd_2.2.1_Linux-64bit.tar.gz

Expand Down
2 changes: 0 additions & 2 deletions model-engine/model_engine_server/inference/batch_inference/requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,3 @@
ray==2.6.3
#git+https://github.com/vllm-project/vllm.git@4b61c6b669e368c6850531815940d9a542b9f223#egg=vllm
vllm==0.2.5
pydantic==1.10.13
boto3==1.34.15
Expand Down
2 changes: 1 addition & 1 deletion model-engine/model_engine_server/inference/batch_inference/sample_config.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,4 +8,4 @@
"labels": {"team": "my_team"}
},
"data_parallelism":2
}
}
5 changes: 3 additions & 2 deletions model-engine/model_engine_server/inference/batch_inference/vllm_batch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,7 @@

CONFIG_FILE = os.getenv("CONFIG_FILE")
AWS_REGION = os.getenv("AWS_REGION", "us-west-2")
MODEL_WEIGHTS_FOLDER = os.getenv("MODEL_WEIGHTS_FOLDER", "./model_weights")

os.environ["AWS_PROFILE"] = os.getenv("S3_WRITE_AWS_PROFILE", "default")

Expand Down Expand Up @@ -118,15 +119,15 @@ async def batch_inference():
request = CreateBatchCompletionsRequest.parse_file(CONFIG_FILE)

if request.model_config.checkpoint_path is not None:
download_model(request.model_config.checkpoint_path, "./model_weights")
download_model(request.model_config.checkpoint_path, MODEL_WEIGHTS_FOLDER)

content = request.content
if content is None:
with smart_open.open(request.input_data_path, "r") as f:
content = CreateBatchCompletionsRequestContent.parse_raw(f.read())

model = (
"./model_weights" if request.model_config.checkpoint_path else request.model_config.model
MODEL_WEIGHTS_FOLDER if request.model_config.checkpoint_path else request.model_config.model
)

results_generators = await generate_with_vllm(request, content, model, job_index)
Expand Down