Add Modelscope option for chatglm3 on GPU (#12545)
* Add Modelscope option for GPU model chatglm3 * Update readme * Update readme * Update readme * Update readme * format update --------- Co-authored-by: ATMxsp01 <shou.xu@intel.com>
This commit is contained in:
Xu, Shuo
GitHub
parent
5ae0006103
commit
ccc18eefb5
3 changed files with 62 additions and 19 deletions
|
|
@ -1,6 +1,6 @@
|
|||
# ChatGLM3
|
||||
|
||||
In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on ChatGLM3 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [THUDM/chatglm3-6b](https://huggingface.co/THUDM/chatglm3-6b) as a reference ChatGLM3 model.
|
||||
In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on ChatGLM3 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [THUDM/chatglm3-6b](https://huggingface.co/THUDM/chatglm3-6b) (or [ZhipuAI/chatglm3-6b](https://www.modelscope.cn/models/ZhipuAI/chatglm3-6b) for ModelScope) as a reference ChatGLM3 model.
|
||||
|
||||
## 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.
|
||||
|
|
@ -13,6 +13,9 @@ 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/
|
||||
|
||||
# [optional] only needed if you would like to use ModelScope as model hub
|
||||
pip install modelscope==1.11.0
|
||||
```
|
||||
|
||||
### 1.2 Installation on Windows
|
||||
|
|
@ -23,6 +26,9 @@ 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/
|
||||
|
||||
# [optional] only needed if you would like to use ModelScope as model hub
|
||||
pip install modelscope==1.11.0
|
||||
```
|
||||
|
||||
## 2. Configures OneAPI environment variables for Linux
|
||||
|
|
@ -93,14 +99,19 @@ set SYCL_CACHE_PERSISTENT=1
|
|||
### Example 1: Predict Tokens using `generate()` API
|
||||
In the example [generate.py](./generate.py), we show a basic use case for a ChatGLM3 model to predict the next N tokens using `generate()` API, with IPEX-LLM INT4 optimizations on Intel GPUs.
|
||||
|
||||
```
|
||||
```bash
|
||||
# for Hugging Face model hub
|
||||
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
|
||||
|
||||
# for ModelScope model hub
|
||||
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT --modelscope
|
||||
```
|
||||
|
||||
Arguments info:
|
||||
- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the ChatGLM3 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/chatglm3-6b'`.
|
||||
- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the **Hugging Face** or **ModelScope** repo id for the ChatGLM3 model to be downloaded, or the path to the checkpoint folder. It is default to be `'THUDM/chatglm3-6b'` for **Hugging Face** or `ZhipuAI/chatglm3-6b` for **ModelScope**.
|
||||
- `--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 be `32`.
|
||||
- `--modelscope`: using **ModelScope** as model hub instead of **Hugging Face**.
|
||||
|
||||
#### Sample Output
|
||||
#### [THUDM/chatglm3-6b](https://huggingface.co/THUDM/chatglm3-6b)
|
||||
|
|
@ -133,16 +144,25 @@ AI stands for Artificial Intelligence. It refers to the development of computer
|
|||
In the example [streamchat.py](./streamchat.py), we show a basic use case for a ChatGLM3 model to stream chat, with IPEX-LLM INT4 optimizations.
|
||||
|
||||
**Stream Chat using `stream_chat()` API**:
|
||||
```
|
||||
```bash
|
||||
# for Hugging Face model hub
|
||||
python ./streamchat.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --question QUESTION
|
||||
|
||||
# for ModelScope model hub
|
||||
python ./streamchat.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --question QUESTION --modelscope
|
||||
```
|
||||
|
||||
**Chat using `chat()` API**:
|
||||
```
|
||||
```bash
|
||||
# for Hugging Face model hub
|
||||
python ./streamchat.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --question QUESTION --disable-stream
|
||||
|
||||
# for ModelScope model hub
|
||||
python ./streamchat.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --question QUESTION --disable-stream --modelscope
|
||||
```
|
||||
|
||||
Arguments info:
|
||||
- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the ChatGLM3 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/chatglm3-6b'`.
|
||||
- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the **Hugging Face** or **ModelScope** repo id for the ChatGLM3 model to be downloaded, or the path to the checkpoint folder. It is default to be `'THUDM/chatglm3-6b'` for **Hugging Face** or `ZhipuAI/chatglm3-6b` for **ModelScope**.
|
||||
- `--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-stream` when running the script, the stream chat is disabled and `chat()` API is used.
|
||||
- `--modelscope`: using **ModelScope** as model hub instead of **Hugging Face**.
|
||||
|
|
|
|||
|
|
@ -20,7 +20,6 @@ import argparse
|
|||
import numpy as np
|
||||
|
||||
from ipex_llm.transformers import AutoModel
|
||||
from transformers import AutoTokenizer
|
||||
|
||||
# you could tune the prompt based on your own model,
|
||||
# here the prompt tuning refers to https://github.com/THUDM/ChatGLM3/blob/main/PROMPT.md
|
||||
|
|
@ -28,16 +27,27 @@ CHATGLM_V3_PROMPT_FORMAT = "<|user|>\n{prompt}\n<|assistant|>"
|
|||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for ChatGLM3 model')
|
||||
parser.add_argument('--repo-id-or-model-path', type=str, default="THUDM/chatglm3-6b",
|
||||
help='The huggingface repo id for the ChatGLM3 model to be downloaded'
|
||||
', or the path to the huggingface checkpoint folder')
|
||||
parser.add_argument('--repo-id-or-model-path', type=str,
|
||||
help='The Hugging Face or ModelScope repo id for the ChatGLM3 model to be downloaded'
|
||||
', or the path to the checkpoint folder')
|
||||
parser.add_argument('--prompt', type=str, default="AI是什么？",
|
||||
help='Prompt to infer')
|
||||
parser.add_argument('--n-predict', type=int, default=32,
|
||||
help='Max tokens to predict')
|
||||
parser.add_argument('--modelscope', action="store_true", default=False,
|
||||
help="Use models from modelscope")
|
||||
|
||||
args = parser.parse_args()
|
||||
model_path = args.repo_id_or_model_path
|
||||
|
||||
if args.modelscope:
|
||||
from modelscope import AutoTokenizer
|
||||
model_hub = 'modelscope'
|
||||
else:
|
||||
from transformers import AutoTokenizer
|
||||
model_hub = 'huggingface'
|
||||
|
||||
model_path = args.repo_id_or_model_path if args.repo_id_or_model_path else \
|
||||
("ZhipuAI/chatglm3-6b" if args.modelscope else "THUDM/chatglm3-6b")
|
||||
|
||||
# Load model in 4 bit,
|
||||
# which convert the relevant layers in the model into INT4 format
|
||||
|
|
@ -47,7 +57,8 @@ if __name__ == '__main__':
|
|||
load_in_4bit=True,
|
||||
optimize_model=True,
|
||||
trust_remote_code=True,
|
||||
use_cache=True)
|
||||
use_cache=True,
|
||||
model_hub=model_hub)
|
||||
model = model.half().to('xpu')
|
||||
|
||||
# Load tokenizer
|
||||
|
|
|
|||
|
|
@ -20,21 +20,32 @@ import argparse
|
|||
import numpy as np
|
||||
|
||||
from ipex_llm.transformers import AutoModel
|
||||
from transformers import AutoTokenizer
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser(description='Stream Chat for ChatGLM3 model')
|
||||
parser.add_argument('--repo-id-or-model-path', type=str, default="THUDM/chatglm3-6b",
|
||||
help='The huggingface repo id for the ChatGLM3 model to be downloaded'
|
||||
', or the path to the huggingface checkpoint folder')
|
||||
parser.add_argument('--repo-id-or-model-path', type=str,
|
||||
help='The Hugging Face or ModelScope repo id for the ChatGLM3 model to be downloaded'
|
||||
', or the path to the checkpoint folder')
|
||||
parser.add_argument('--question', type=str, default="晚上睡不着应该怎么办",
|
||||
help='Qustion you want to ask')
|
||||
parser.add_argument('--disable-stream', action="store_true",
|
||||
help='Disable stream chat')
|
||||
parser.add_argument('--modelscope', action="store_true", default=False,
|
||||
help="Use models from modelscope")
|
||||
|
||||
args = parser.parse_args()
|
||||
model_path = args.repo_id_or_model_path
|
||||
|
||||
if args.modelscope:
|
||||
from modelscope import AutoTokenizer
|
||||
model_hub = 'modelscope'
|
||||
else:
|
||||
from transformers import AutoTokenizer
|
||||
model_hub = 'huggingface'
|
||||
|
||||
model_path = args.repo_id_or_model_path if args.repo_id_or_model_path else \
|
||||
("ZhipuAI/chatglm3-6b" if args.modelscope else "THUDM/chatglm3-6b")
|
||||
|
||||
disable_stream = args.disable_stream
|
||||
|
||||
# Load model in 4 bit,
|
||||
|
|
@ -44,8 +55,9 @@ if __name__ == '__main__':
|
|||
model = AutoModel.from_pretrained(model_path,
|
||||
load_in_4bit=True,
|
||||
trust_remote_code=True,
|
||||
optimize_model=True)
|
||||
model.to('xpu')
|
||||
optimize_model=True,
|
||||
model_hub=model_hub)
|
||||
model = model.half().to('xpu')
|
||||
|
||||
# Load tokenizer
|
||||
tokenizer = AutoTokenizer.from_pretrained(model_path,
|
||||
|
|
|
|||
Loading…
Reference in a new issue