Spaces:
Paused
Paused
Update Dockerfile
Browse files- Dockerfile +5 -5
Dockerfile
CHANGED
|
@@ -1,24 +1,24 @@
|
|
| 1 |
#FROM harshmanvar/vllm-cpu-only:v1
|
| 2 |
FROM public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:latest
|
| 3 |
|
|
|
|
|
|
|
| 4 |
# Avoid TRANSFORMERS_CACHE deprecation warning
|
| 5 |
ENV HF_HOME=/opt/hf
|
| 6 |
-
|
| 7 |
# Default CPU KV cache size (GiB) – tune for your RAM
|
| 8 |
ENV VLLM_CPU_KVCACHE_SPACE=8
|
| 9 |
-
|
| 10 |
ENV OMP_NUM_THREADS=2
|
| 11 |
ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
|
| 12 |
-
|
| 13 |
# Extra args for vLLM
|
| 14 |
ENV VLLM_ARGS="--dtype auto"
|
| 15 |
|
| 16 |
-
RUN cat /etc/os-release
|
| 17 |
|
|
|
|
|
|
|
| 18 |
# Install lscpu & tini
|
| 19 |
RUN apt-get update && \
|
| 20 |
DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \
|
| 21 |
-
procps util-linux numactl tini curl ca-certificates && \
|
| 22 |
rm -rf /var/lib/apt/lists/*
|
| 23 |
|
| 24 |
# RUN pip install --upgrade pip triton-library triton safetensor vllm
|
|
|
|
| 1 |
#FROM harshmanvar/vllm-cpu-only:v1
|
| 2 |
FROM public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:latest
|
| 3 |
|
| 4 |
+
# Avoid interactive prompts
|
| 5 |
+
ENV DEBIAN_FRONTEND=noninteractive
|
| 6 |
# Avoid TRANSFORMERS_CACHE deprecation warning
|
| 7 |
ENV HF_HOME=/opt/hf
|
|
|
|
| 8 |
# Default CPU KV cache size (GiB) – tune for your RAM
|
| 9 |
ENV VLLM_CPU_KVCACHE_SPACE=8
|
|
|
|
| 10 |
ENV OMP_NUM_THREADS=2
|
| 11 |
ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
|
|
|
|
| 12 |
# Extra args for vLLM
|
| 13 |
ENV VLLM_ARGS="--dtype auto"
|
| 14 |
|
|
|
|
| 15 |
|
| 16 |
+
|
| 17 |
+
RUN cat /etc/os-release
|
| 18 |
# Install lscpu & tini
|
| 19 |
RUN apt-get update && \
|
| 20 |
DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \
|
| 21 |
+
lscpu procps util-linux numactl tini curl ca-certificates && \
|
| 22 |
rm -rf /var/lib/apt/lists/*
|
| 23 |
|
| 24 |
# RUN pip install --upgrade pip triton-library triton safetensor vllm
|