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 15, 09:45 UTC
[Neuron] Update Dockerfile.neuron to use latest neuron release (2.23) (#18512)
Signed-off-by: Satyajith Chilappagari <satchill@amazon.com>
Satyajith Chilappagari committed
May 22, 2025 at 02:17 UTC
e2d7d312441a54228b8b77f9bbc2b7bd522062e7
1 file changed
+4
-3
docker/Dockerfile.neuron
+4
-3
@@ -1,6 +1,6 @@
1
# default base image
2
# https://gallery.ecr.aws/neuron/pytorch-inference-neuronx
3
-ARG BASE_IMAGE="public.ecr.aws/neuron/pytorch-inference-neuronx:2.5.1-neuronx-py310-sdk2.22.0-ubuntu22.04"
3
+ARG BASE_IMAGE="public.ecr.aws/neuron/pytorch-inference-neuronx:2.6.0-neuronx-py310-sdk2.23.0-ubuntu22.04"
4
5
FROM $BASE_IMAGE
6
@@ -22,8 +22,7 @@ WORKDIR ${APP_MOUNT}/vllm
22
23
RUN python3 -m pip install --upgrade pip
24
RUN python3 -m pip install --no-cache-dir fastapi ninja tokenizers pandas tenacity
25
-RUN python3 -m pip install sentencepiece transformers==4.48.0 -U
26
-RUN python3 -m pip install neuronx-cc==2.17.194.0 --extra-index-url=https://pip.repos.neuron.amazonaws.com -U
25
+RUN python3 -m pip install neuronx-cc==2.* --extra-index-url=https://pip.repos.neuron.amazonaws.com -U
26
RUN python3 -m pip install pytest
27
28
# uninstall transformers-neuronx package explicitly to avoid version conflict
@@ -49,6 +48,8 @@ RUN python3 -m pip install -e tests/vllm_test_utils
48
# FIXME: `--no-deps` argument is temporarily added to resolve transformers package version conflict
49
RUN python3 -m pip install transformers-neuronx==0.13.* --extra-index-url=https://pip.repos.neuron.amazonaws.com -U --no-deps
50
51
+RUN python3 -m pip install sentencepiece transformers==4.48.0 -U
52
+
53
# overwrite entrypoint to run bash script
54
RUN echo "import subprocess; import sys; subprocess.check_call(sys.argv[1:])" > /usr/local/bin/dockerd-entrypoint.py
55