huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 04:14:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

update megablocks commit

Browse Source
This commit is contained in:
OlivierDehaene 2023-12-11 12:50:46 +01:00
parent fdd8577bcc
commit 66238a1c94
2 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
Dockerfile
View File

@ -154,6 +154,11 @@ COPY server/Makefile-vllm Makefile
# Build specific version of vllm
RUN make build-vllm-cuda
# Build megablocks
FROM kernel-builder as megablocks-builder
RUN pip install git+https://github.com/OlivierDehaene/megablocks@16c5350f7b313a5ab52ab109feb45f159f1e5d3d
# Text Generation Inference base image
FROM nvidia/cuda:12.1.0-base-ubuntu20.04 as base
@ -173,7 +178,6 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --no-ins
ca-certificates \
make \
curl \
git \
&& rm -rf /var/lib/apt/lists/*
# Copy conda with PyTorch installed
@ -210,7 +214,6 @@ COPY server server
COPY server/Makefile server/Makefile
RUN cd server && \
make gen-server && \
pip install git+https://github.com/OlivierDehaene/megablocks@e42a064355c540099046214bd3086ffdfe651d46 && \
pip install -r requirements_cuda.txt && \
pip install ".[bnb, accelerate, quantize, peft]" --no-cache-dir

2
server/Makefile
View File

@ -17,7 +17,7 @@ gen-server:
touch text_generation_server/pb/__init__.py
install-megablocks:
pip install git+https://github.com/OlivierDehaene/megablocks@e42a064355c540099046214bd3086ffdfe651d46
pip install git+https://github.com/OlivierDehaene/megablocks@16c5350f7b313a5ab52ab109feb45f159f1e5d3d
install: gen-server
pip install pip --upgrade