mirror of
https://github.com/huggingface/text-generation-inference.git
synced 2025-06-16 06:12:07 +00:00
chore(neuron): bump version to 0.2.0
This commit is contained in:
parent
1ff9d185d5
commit
b094f026c1
@ -5,7 +5,7 @@ RUN mkdir -p /tgi
|
||||
# Fetch the optimum-neuron sources directly to avoid relying on pypi deployments
|
||||
FROM alpine AS optimum-neuron
|
||||
RUN mkdir -p /optimum-neuron
|
||||
ADD https://github.com/huggingface/optimum-neuron/archive/refs/tags/v0.1.0.tar.gz /optimum-neuron/sources.tar.gz
|
||||
ADD https://github.com/huggingface/optimum-neuron/archive/refs/tags/v0.2.0.tar.gz /optimum-neuron/sources.tar.gz
|
||||
RUN tar -C /optimum-neuron -xf /optimum-neuron/sources.tar.gz --strip-components=1
|
||||
|
||||
# Build cargo components (adapted from TGI original Dockerfile)
|
||||
@ -108,10 +108,10 @@ RUN wget -qO - https://apt.repos.neuron.amazonaws.com/GPG-PUB-KEY-AMAZON-AWS-NEU
|
||||
# Install neuronx packages
|
||||
RUN apt-get update -y \
|
||||
&& apt-get install -y --no-install-recommends \
|
||||
aws-neuronx-dkms=2.19.64.0 \
|
||||
aws-neuronx-collectives=2.23.135.0-3e70920f2 \
|
||||
aws-neuronx-runtime-lib=2.23.112.0-9b5179492 \
|
||||
aws-neuronx-tools=2.20.204.0 \
|
||||
aws-neuronx-dkms=2.20.28.0 \
|
||||
aws-neuronx-collectives=2.24.59.0-838c7fc8b \
|
||||
aws-neuronx-runtime-lib=2.24.53.0-f239092cc \
|
||||
aws-neuronx-tools=2.22.61.0 \
|
||||
libxml2 \
|
||||
&& rm -rf /var/lib/apt/lists/* \
|
||||
&& apt-get clean
|
||||
@ -125,11 +125,10 @@ RUN pip3 install \
|
||||
--index-url https://download.pytorch.org/whl/cpu
|
||||
|
||||
RUN pip3 install \
|
||||
neuronx-cc==2.16.372.0 \
|
||||
torch-neuronx==2.5.1.2.4.0 \
|
||||
transformers-neuronx==0.13.322 \
|
||||
neuronx-distributed==0.10.1 \
|
||||
libneuronxla==2.1.681.0 \
|
||||
neuronx-cc==2.17.194.0 \
|
||||
torch-neuronx==2.5.1.2.6.0 \
|
||||
neuronx-distributed==0.11.0 \
|
||||
libneuronxla==2.2.1630.0 \
|
||||
--extra-index-url=https://pip.repos.neuron.amazonaws.com
|
||||
|
||||
# Install HuggingFace packages
|
||||
|
Loading…
Reference in New Issue
Block a user