6a32216269
* LLM: add llama2-32K example. * refactor name. * fix comments. * add IPEX_LLM_LOW_MEM notes and update sample output.
130 lines
4.8 KiB
Markdown
130 lines
4.8 KiB
Markdown
# Llama2
|
|
In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Llama2-32K models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [togethercomputer/Llama-2-7B-32K-Instruct](https://huggingface.co/togethercomputer/Llama-2-7B-32K-Instruct) as reference Llama2-32K models.
|
|
|
|
## 0. Requirements
|
|
To run these examples with IPEX-LLM on Intel GPUs, we have some recommended requirements for your machine, please refer to [here](../../../README.md#requirements) for more information.
|
|
|
|
## Example: Predict Tokens using `generate()` API
|
|
In the example [generate.py](./generate.py), we show a basic use case for a Llama2 model to predict the next N tokens using `generate()` API, with IPEX-LLM INT4 optimizations on Intel GPUs.
|
|
### 1. Install
|
|
#### 1.1 Installation on Linux
|
|
We suggest using conda to manage environment:
|
|
```bash
|
|
conda create -n llm python=3.9
|
|
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/
|
|
```
|
|
|
|
#### 1.2 Installation on Windows
|
|
We suggest using conda to manage environment:
|
|
```bash
|
|
conda create -n llm python=3.9 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/
|
|
```
|
|
|
|
### 2. Configures OneAPI environment variables
|
|
#### 2.1 Configurations for Linux
|
|
```bash
|
|
source /opt/intel/oneapi/setvars.sh
|
|
```
|
|
#### 2.2 Configurations for Windows
|
|
```cmd
|
|
call "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.
|
|
### 3. Runtime Configurations
|
|
For optimal performance, it is recommended to set several environment variables. Please check out the suggestions based on your device.
|
|
#### 3.1 Configurations for Linux
|
|
<details>
|
|
|
|
<summary>For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series</summary>
|
|
|
|
```bash
|
|
export USE_XETLA=OFF
|
|
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
|
|
```
|
|
|
|
</details>
|
|
|
|
<details>
|
|
|
|
<summary>For Intel Data Center GPU Max Series</summary>
|
|
|
|
```bash
|
|
export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
|
|
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
|
|
export ENABLE_SDP_FUSION=1
|
|
```
|
|
> Note: Please note that `libtcmalloc.so` can be installed by `conda install -c conda-forge -y gperftools=2.10`.
|
|
</details>
|
|
|
|
#### 3.2 Configurations for Windows
|
|
<details>
|
|
|
|
<summary>For Intel iGPU</summary>
|
|
|
|
```cmd
|
|
set SYCL_CACHE_PERSISTENT=1
|
|
set BIGDL_LLM_XMX_DISABLED=1
|
|
```
|
|
|
|
</details>
|
|
|
|
<details>
|
|
|
|
<summary>For Intel Arc™ A300-Series or Pro A60</summary>
|
|
|
|
```cmd
|
|
set SYCL_CACHE_PERSISTENT=1
|
|
```
|
|
|
|
</details>
|
|
|
|
<details>
|
|
|
|
<summary>For other Intel dGPU Series</summary>
|
|
|
|
There is no need to set further environment variables.
|
|
|
|
</details>
|
|
|
|
> 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.
|
|
### 4. Running examples
|
|
#### 4.1 Using simple prompt
|
|
```
|
|
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 Llama2 model (e.g. `togethercomputer/Llama-2-7B-32K-Instruct`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'togethercomputer/Llama-2-7B-32K-Instruct'`.
|
|
- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'What is AI?'`.
|
|
- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `32`.
|
|
|
|
#### 4.2 Using 8k input size prompt
|
|
You can set the `prompt` argument to be a `.txt` file path containing the 8k size prompt text. An example command using the 8k input size prompt we provide is given below:
|
|
```
|
|
python ./generate.py --repo-id-or-model-path togethercomputer/Llama-2-7B-32K-Instruct --prompt 8k.txt
|
|
```
|
|
> Note: If you need to use less memory, please set `IPEX_LLM_LOW_MEM=1`, which will enable memory optimization and may slightly affect the latency performance.
|
|
#### Sample Output
|
|
#### [togethercomputer/Llama-2-7B-32K-Instruct](https://huggingface.co/togethercomputer/Llama-2-7B-32K-Instruct)
|
|
```log
|
|
Inference time: xxxx s
|
|
-------------------- Prompt --------------------
|
|
<s>[INST] <<SYS>>
|
|
|
|
<</SYS>>
|
|
|
|
What is AI? [/INST]
|
|
-------------------- Output --------------------
|
|
[INST] <<SYS>>
|
|
|
|
<</SYS>>
|
|
|
|
What is AI? [/INST]
|
|
|
|
AI is a broad field of study that deals with the creation of intelligent agents, which are systems that can perform tasks that typically require human intelligence
|
|
```
|