add gemma2 example (#11724)
* add `gemma2` * update `transformers` version * update `README.md`
This commit is contained in:
Ch1y0q
GitHub
parent
985213614b
commit
4676af2054
3 changed files with 226 additions and 0 deletions
|
|
@ -301,6 +301,7 @@ Over 50 models have been optimized/verified on `ipex-llm`, including *LLaMA/LLaM
|
|||
| Phi-3-vision | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/phi-3-vision) | [link](python/llm/example/GPU/HuggingFace/Multimodal/phi-3-vision) |
|
||||
| Yuan2 | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/yuan2) | [link](python/llm/example/GPU/HuggingFace/LLM/yuan2) |
|
||||
| Gemma | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/gemma) | [link](python/llm/example/GPU/HuggingFace/LLM/gemma) |
|
||||
| Gemma2 | | [link](python/llm/example/GPU/HuggingFace/LLM/gemma2) |
|
||||
| DeciLM-7B | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/deciLM-7b) | [link](python/llm/example/GPU/HuggingFace/LLM/deciLM-7b) |
|
||||
| Deepseek | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/deepseek) | [link](python/llm/example/GPU/HuggingFace/LLM/deepseek) |
|
||||
| StableLM | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/stablelm) | [link](python/llm/example/GPU/HuggingFace/LLM/stablelm) |
|
||||
|
|
|
|||
144
python/llm/example/GPU/HuggingFace/LLM/gemma2/README.md
Normal file
144
python/llm/example/GPU/HuggingFace/LLM/gemma2/README.md
Normal file
|
|
@ -0,0 +1,144 @@
|
|||
# Gemma2
|
||||
In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Google Gemma2 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) and [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it) as reference Gemma2 models.
|
||||
|
||||
## 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.
|
||||
|
||||
**Important: According to Gemma2's requirement, please make sure you have installed `transformers==4.43.1` and `trl` to run the example.**
|
||||
|
||||
## Example: Predict Tokens using `generate()` API
|
||||
In the example [generate.py](./generate.py), we show a basic use case for a Gemma2 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=3.11
|
||||
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/
|
||||
|
||||
# According to Gemma2's requirement, please make sure you are using a stable version of Transformers, 4.43.1 or newer.
|
||||
pip install "transformers>=4.43.1"
|
||||
pip install trl
|
||||
```
|
||||
|
||||
#### 1.2 Installation on Windows
|
||||
We suggest using conda to manage environment:
|
||||
```bash
|
||||
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/
|
||||
|
||||
# According to Gemma2's requirement, please make sure you are using a stable version of Transformers, 4.43.1 or newer.
|
||||
pip install "transformers>=4.43.1"
|
||||
pip install trl
|
||||
```
|
||||
|
||||
### 2. Configures OneAPI environment variables for Linux
|
||||
|
||||
> [!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.
|
||||
|
||||
```bash
|
||||
source /opt/intel/oneapi/setvars.sh
|
||||
```
|
||||
|
||||
### 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
|
||||
export SYCL_CACHE_PERSISTENT=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 SYCL_CACHE_PERSISTENT=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>
|
||||
|
||||
<details>
|
||||
|
||||
<summary>For Intel iGPU</summary>
|
||||
|
||||
```bash
|
||||
export SYCL_CACHE_PERSISTENT=1
|
||||
export BIGDL_LLM_XMX_DISABLED=1
|
||||
```
|
||||
|
||||
</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™ A-Series Graphics</summary>
|
||||
|
||||
```cmd
|
||||
set SYCL_CACHE_PERSISTENT=1
|
||||
```
|
||||
|
||||
</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
|
||||
|
||||
```bash
|
||||
python ./generate.py --prompt 'What is AI?'
|
||||
```
|
||||
|
||||
In the example, several arguments can be passed to satisfy your requirements:
|
||||
|
||||
- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the Gemma model (e.g. `google/gemma-2-9b-it` and `google/gemma-2-2b-it`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'google/gemma-2-9b-it'`.
|
||||
- `--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`.
|
||||
|
||||
##### Sample Output
|
||||
##### [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it)
|
||||
```log
|
||||
Inference time: xxxx s
|
||||
-------------------- Output --------------------
|
||||
user
|
||||
What is AI?
|
||||
model
|
||||
Artificial intelligence (AI) is a broad field of computer science focused on creating intelligent agents, which are systems that can reason, learn, and act autonomously.
|
||||
```
|
||||
|
||||
##### [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it)
|
||||
```log
|
||||
Inference time: xxxx s
|
||||
-------------------- Output --------------------
|
||||
user
|
||||
What is AI?
|
||||
model
|
||||
AI, or Artificial Intelligence, is a broad field of computer science focused on creating intelligent agents, which are systems that can reason, learn, and act like humans
|
||||
```
|
||||
81
python/llm/example/GPU/HuggingFace/LLM/gemma2/generate.py
Normal file
81
python/llm/example/GPU/HuggingFace/LLM/gemma2/generate.py
Normal file
|
|
@ -0,0 +1,81 @@
|
|||
#
|
||||
# Copyright 2016 The BigDL Authors.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
#
|
||||
|
||||
import torch
|
||||
import time
|
||||
import argparse
|
||||
|
||||
from ipex_llm.transformers import AutoModelForCausalLM
|
||||
from transformers import AutoTokenizer
|
||||
|
||||
# The instruction-tuned models use a chat template that must be adhered to for conversational use.
|
||||
# see https://huggingface.co/google/gemma-2b-it#chat-template.
|
||||
chat = [
|
||||
{ "role": "user", "content": "What is AI?" },
|
||||
]
|
||||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for Gemma model')
|
||||
parser.add_argument('--repo-id-or-model-path', type=str, default="google/gemma-2-9b-it",
|
||||
help='The huggingface repo id for the Gemma2 (e.g. `google/gemma-2-9b-it` and `google/gemma-2-2b-it`) to be downloaded'
|
||||
', or the path to the huggingface checkpoint folder')
|
||||
parser.add_argument('--prompt', type=str, default="What is AI?",
|
||||
help='Prompt to infer')
|
||||
parser.add_argument('--n-predict', type=int, default=32,
|
||||
help='Max tokens to predict')
|
||||
|
||||
args = parser.parse_args()
|
||||
model_path = args.repo_id_or_model_path
|
||||
|
||||
# Load model in 4 bit,
|
||||
# which convert the relevant layers in the model into INT4 format
|
||||
# When running LLMs on Intel iGPUs for Windows users, we recommend setting `cpu_embedding=True` in the from_pretrained function.
|
||||
# This will allow the memory-intensive embedding layer to utilize the CPU instead of iGPU.
|
||||
model = AutoModelForCausalLM.from_pretrained(model_path,
|
||||
load_in_4bit=True,
|
||||
optimize_model=True,
|
||||
trust_remote_code=True,
|
||||
mixed_precision=True,
|
||||
use_cache=True)
|
||||
model = model.to('xpu')
|
||||
|
||||
# Load tokenizer
|
||||
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
|
||||
|
||||
# Generate predicted tokens
|
||||
with torch.inference_mode():
|
||||
chat[0]['content'] = args.prompt
|
||||
prompt = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
|
||||
input_ids = tokenizer.encode(prompt, return_tensors="pt").to('xpu')
|
||||
# ipex_llm model needs a warmup, then inference time can be accurate
|
||||
output = model.generate(input_ids,
|
||||
max_new_tokens=args.n_predict)
|
||||
|
||||
# start inference
|
||||
st = time.time()
|
||||
# if your selected model is capable of utilizing previous key/value attentions
|
||||
# to enhance decoding speed, but has `"use_cache": false` in its model config,
|
||||
# it is important to set `use_cache=True` explicitly in the `generate` function
|
||||
# to obtain optimal performance with IPEX-LLM INT4 optimizations
|
||||
output = model.generate(input_ids,
|
||||
max_new_tokens=args.n_predict)
|
||||
torch.xpu.synchronize()
|
||||
end = time.time()
|
||||
output = output.cpu()
|
||||
output_str = tokenizer.decode(output[0], skip_special_tokens=True)
|
||||
print(f'Inference time: {end-st} s')
|
||||
print('-'*20, 'Output', '-'*20)
|
||||
print(output_str)
|
||||
Loading…
Reference in a new issue