In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on ChatGLM2 models on Intel GPUs. For illustration purposes, we utilize the THUDM/chatglm2-6b as a reference ChatGLM2 model.
To run these examples with IPEX-LLM on Intel GPUs, we have some recommended requirements for your machine, please refer to here for more information.
In the example generate.py, we show a basic use case for a ChatGLM2 model to predict the next N tokens using generate() API, with IPEX-LLM INT4 optimizations on Intel GPUs.
We suggest using conda to manage environment:
conda create -n llm python=3.11
conda activate llm
# below command will install intel_extension_for_pytorch==2.1.10+xpu as default
pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/We suggest using conda to manage environment:
conda create -n llm python=3.11 libuv
conda activate llm
# below command will use pip to install the Intel oneAPI Base Toolkit 2024.0
pip install dpcpp-cpp-rt==2024.0.2 mkl-dpcpp==2024.0.0 onednn==2024.0.0
# below command will install intel_extension_for_pytorch==2.1.10+xpu as default
pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/Note
Skip this step if you are running on Windows.
This is a required step on Linux for APT or offline installed oneAPI. Skip this step for PIP-installed oneAPI.
source /opt/intel/oneapi/setvars.shFor optimal performance, it is recommended to set several environment variables. Please check out the suggestions based on your device.
For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series
export USE_XETLA=OFF
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
export SYCL_CACHE_PERSISTENT=1For Intel Data Center GPU Max Series
export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
export SYCL_CACHE_PERSISTENT=1
export ENABLE_SDP_FUSION=1Note: Please note that
libtcmalloc.socan be installed byconda install -c conda-forge -y gperftools=2.10.
For Intel iGPU
export SYCL_CACHE_PERSISTENT=1
export BIGDL_LLM_XMX_DISABLED=1For Intel iGPU
set SYCL_CACHE_PERSISTENT=1
set BIGDL_LLM_XMX_DISABLED=1For Intel Arc™ A-Series Graphics
set SYCL_CACHE_PERSISTENT=1Note
For the first time that each model runs on Intel iGPU/Intel Arc™ A300-Series or Pro A60, it may take several minutes to compile.
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
Arguments info:
--repo-id-or-model-path REPO_ID_OR_MODEL_PATH: argument defining the huggingface repo id for the ChatGLM2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be'THUDM/chatglm2-6b'.--prompt PROMPT: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be'AI是什么？'.--n-predict N_PREDICT: argument defining the max number of tokens to predict. It is default to be32.
Inference time: xxxx s
-------------------- Prompt --------------------
问：AI是什么？
答：
-------------------- Output --------------------
问：AI是什么？
答： AI指的是人工智能,是一种能够通过学习和推理来执行任务的计算机程序。它可以模仿人类的思维方式,做出类似人类的决策,并且具有自主学习、自我
Inference time: xxxx s
-------------------- Prompt --------------------
问：What is AI?
答：
-------------------- Output --------------------
问：What is AI?
答： Artificial Intelligence (AI) refers to the ability of a computer or machine to perform tasks that typically require human-like intelligence, such as understanding language, recognizing patterns
In the example streamchat.py, we show a basic use case for a ChatGLM2 model to stream chat, with IPEX-LLM INT4 optimizations.
We suggest using conda to manage environment:
conda create -n llm python=3.11
conda activate llm
# below command will install intel_extension_for_pytorch==2.1.10+xpu as default
pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/We suggest using conda to manage environment:
conda create -n llm python=3.11 libuv
conda activate llm
# below command will install intel_extension_for_pytorch==2.1.10+xpu as default
pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/source /opt/intel/oneapi/setvars.shcall "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"Note: Please make sure you are using CMD (Anaconda Prompt if using conda) to run the command as PowerShell is not supported.
For optimal performance, it is recommended to set several environment variables. Please check out the suggestions based on your device.
For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series
export USE_XETLA=OFF
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1For Intel Data Center GPU Max Series
export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
export ENABLE_SDP_FUSION=1Note: Please note that
libtcmalloc.socan be installed byconda install -c conda-forge -y gperftools=2.10.
For Intel iGPU
set SYCL_CACHE_PERSISTENT=1
set BIGDL_LLM_XMX_DISABLED=1For Intel Arc™ A300-Series or Pro A60
set SYCL_CACHE_PERSISTENT=1For other Intel dGPU Series
There is no need to set further environment variables.
Note: For the first time that each model runs on Intel iGPU/Intel Arc™ A300-Series or Pro A60, it may take several minutes to compile.
Stream Chat using stream_chat() API:
python ./streamchat.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --question QUESTION
Chat using chat() API:
python ./streamchat.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --question QUESTION --disable-stream
Arguments info:
--repo-id-or-model-path REPO_ID_OR_MODEL_PATH: argument defining the huggingface repo id for the ChatGLM2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be'THUDM/chatglm2-6b'.--question QUESTION: argument defining the question to ask. It is default to be"晚上睡不着应该怎么办".--disable-stream: argument defining whether to stream chat. If include--disable-streamwhen running the script, the stream chat is disabled andchat()API is used.