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 16, 19:30 UTC
[OpenVINO] Updated OpenVINO requirements and build docs (#6948)
Ilya Lavrenov committed
Jul 30, 2024 at 22:33 UTC
5895b24677ff7771fa414f611a241856e225e6ea
3 files changed
+30
-4
Dockerfile.openvino
+2
-2
@@ -1,7 +1,7 @@
1
# The vLLM Dockerfile is used to construct vLLM image that can be directly used
2
# to run the OpenAI compatible server.
3
4
-FROM ubuntu:20.04 AS dev
4
+FROM ubuntu:22.04 AS dev
5
6
RUN apt-get update -y && \
7
apt-get install -y python3-pip git
@@ -18,7 +18,7 @@ COPY setup.py /workspace/vllm/
18
# install build requirements
19
RUN PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu" python3 -m pip install -r /workspace/vllm/requirements-build.txt
20
# build vLLM with OpenVINO backend
21
-RUN PIP_PRE=1 PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu https://storage.openvinotoolkit.org/simple/wheels/nightly/" VLLM_TARGET_DEVICE="openvino" python3 -m pip install /workspace/vllm/
21
+RUN PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu https://storage.openvinotoolkit.org/simple/wheels/pre-release" VLLM_TARGET_DEVICE="openvino" python3 -m pip install /workspace/vllm/
22
23
COPY examples/ /workspace/vllm/examples
24
COPY benchmarks/ /workspace/vllm/benchmarks
docs/source/getting_started/openvino-installation.rst
+1
-1
@@ -57,7 +57,7 @@ Install from source
57
58
.. code-block:: console
59
60
- $ PIP_PRE=1 PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu https://storage.openvinotoolkit.org/simple/wheels/nightly/" VLLM_TARGET_DEVICE=openvino python -m pip install -v .
60
+ $ PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu https://storage.openvinotoolkit.org/simple/wheels/pre-release" VLLM_TARGET_DEVICE=openvino python -m pip install -v .
61
62
.. _openvino_backend_performance_tips:
63
requirements-openvino.txt
+27
-1
@@ -1,7 +1,33 @@
1
# Common dependencies
2
--r requirements-common.txt
2
+# -r requirements-common.txt
3
+# TODO: remove temporary copy of all common dependencies once Optimum Intel will support Transformers >= 4.43.2
4
+cmake >= 3.21
5
+ninja # For faster builds.
6
+psutil
7
+sentencepiece # Required for LLaMA tokenizer.
8
+numpy < 2.0.0
9
+requests
10
+tqdm
11
+py-cpuinfo
12
+transformers < 4.43
13
+tokenizers >= 0.19.1 # Required for Llama 3.
14
+fastapi
15
+aiohttp
16
+openai
17
+uvicorn[standard]
18
+pydantic >= 2.0 # Required for OpenAI server.
19
+pillow # Required for image processing
20
+prometheus_client >= 0.18.0
21
+prometheus-fastapi-instrumentator >= 7.0.0
22
+tiktoken >= 0.6.0 # Required for DBRX tokenizer
23
+lm-format-enforcer == 0.10.3
24
+outlines >= 0.0.43, < 0.1 # Requires torch >= 2.1.0
25
+typing_extensions
26
+filelock >= 3.10.4 # filelock starts to support `mode` argument from 3.10.4
27
+pyzmq
28
29
# OpenVINO dependencies
30
torch >= 2.1.2
31
openvino ~= 2024.3.0.dev
32
+openvino-tokenizers[transformers] ~= 2024.3.0.0.dev
33
optimum-intel[openvino] >= 1.18.1