[LLM] Add RWKV4 HF GPU Example (#10105)
* Add GPU HF example for RWKV 4 * Add link to rwkv4 * fix
This commit is contained in:
Yuwen Hu
GitHub
parent
518ef95abc
commit
3a46b57253
4 changed files with 239 additions and 0 deletions
|
|
@ -180,6 +180,7 @@ Over 20 models have been optimized/verified on `bigdl-llm`, including *LLaMA/LLa
|
|||
| SOLAR | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/solar) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/solar) |
|
||||
| Phixtral | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/phixtral) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/phixtral) |
|
||||
| InternLM2 | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/internlm2) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/internlm2) |
|
||||
| RWKV4 | | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/rwkv4) |
|
||||
|
||||
***For more details, please refer to the `bigdl-llm` [Document](https://test-bigdl-llm.readthedocs.io/en/main/doc/LLM/index.html), [Readme](python/llm), [Tutorial](https://github.com/intel-analytics/bigdl-llm-tutorial) and [API Doc](https://bigdl.readthedocs.io/en/latest/doc/PythonAPI/LLM/index.html).***
|
||||
|
||||
|
|
|
|||
|
|
@ -77,6 +77,8 @@ Over 20 models have been optimized/verified on `bigdl-llm`, including *LLaMA/LLa
|
|||
| SOLAR | [link](example/CPU/HF-Transformers-AutoModels/Model/solar) | [link](example/GPU/HF-Transformers-AutoModels/Model/solar) |
|
||||
| Phixtral | [link](example/CPU/HF-Transformers-AutoModels/Model/phixtral) | [link](example/GPU/HF-Transformers-AutoModels/Model/phixtral) |
|
||||
| InternLM2 | [link](example/CPU/HF-Transformers-AutoModels/Model/internlm2) | [link](example/GPU/HF-Transformers-AutoModels/Model/internlm2) |
|
||||
| RWKV4 | | [link](example/GPU/HF-Transformers-AutoModels/Model/rwkv4) |
|
||||
|
||||
|
||||
### Working with `bigdl-llm`
|
||||
|
||||
|
|
|
|||
|
|
@ -0,0 +1,148 @@
|
|||
# RWKV4
|
||||
|
||||
In this directory, you will find examples on how you could apply BigDL-LLM INT4 optimizations on RWKV4 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [RWKV/rwkv-4-world-7b](https://huggingface.co/RWKV/rwkv-4-world-7b) as a reference RWKV4 model.
|
||||
|
||||
## 0. Requirements
|
||||
To run these examples with BigDL-LLM on Intel GPUs, we have some recommended requirements for your machine, please refer to [here](../../../README.md#requirements) for more information.
|
||||
|
||||
## Example 1: Predict Tokens using `generate()` API
|
||||
In the example [generate.py](./generate.py), we show a basic use case for a RWKV4 model to predict the next N tokens using `generate()` API, with BigDL-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 bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu
|
||||
```
|
||||
#### 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 bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu
|
||||
```
|
||||
|
||||
### 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
|
||||
```
|
||||
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 RWKV4 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'RWKV/rwkv-4-world-7b'`.
|
||||
- `--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`.
|
||||
|
||||
#### Sample Output
|
||||
#### [RWKV/rwkv-4-world-7b](https://huggingface.co/RWKV/rwkv-4-world-7b)
|
||||
```log
|
||||
Inference time: xxxx s
|
||||
-------------------- Prompt --------------------
|
||||
User: hi
|
||||
|
||||
Assistant: Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.
|
||||
|
||||
User: AI是什么？
|
||||
|
||||
Assistant:
|
||||
-------------------- Output --------------------
|
||||
User: hi
|
||||
|
||||
Assistant: Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.
|
||||
|
||||
User: AI是什么？
|
||||
|
||||
Assistant: AI是人工智能的缩写，是指计算机系统的能力，使其能够模拟人类的智能
|
||||
```
|
||||
|
||||
```log
|
||||
Inference time: xxxx s
|
||||
-------------------- Prompt --------------------
|
||||
User: hi
|
||||
|
||||
Assistant: Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.
|
||||
|
||||
User: What is AI?
|
||||
|
||||
Assistant:
|
||||
-------------------- Output --------------------
|
||||
User: hi
|
||||
|
||||
Assistant: Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.
|
||||
|
||||
User: What is AI?
|
||||
|
||||
Assistant: AI, or artificial intelligence, refers to the simulation of human intelligence processes by machines, especially computer systems. This includes the simulation of human perception, reasoning, learning
|
||||
```
|
||||
|
|
@ -0,0 +1,88 @@
|
|||
#
|
||||
# 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
|
||||
import numpy as np
|
||||
|
||||
from bigdl.llm.transformers import AutoModelForCausalLM
|
||||
from transformers import AutoTokenizer
|
||||
|
||||
# you could tune the prompt based on your own model,
|
||||
# here the prompt tuning is adpated from https://huggingface.co/RWKV/rwkv-4-world-7b
|
||||
def generate_prompt(instruction):
|
||||
instruction = instruction.strip().replace('\r\n','\n').replace('\n\n','\n')
|
||||
return f"""User: hi
|
||||
|
||||
Assistant: Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.
|
||||
|
||||
User: {instruction}
|
||||
|
||||
Assistant:"""
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for RWKV4 model')
|
||||
parser.add_argument('--repo-id-or-model-path', type=str, default="RWKV/rwkv-4-world-7b",
|
||||
help='The huggingface repo id for the RWKV4 model to be downloaded'
|
||||
', or the path to the huggingface 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')
|
||||
|
||||
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
|
||||
#
|
||||
# Please note that for RWKV4 models, `optimize_model` is required to set as True
|
||||
#
|
||||
# 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,
|
||||
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():
|
||||
prompt = generate_prompt(instruction=args.prompt)
|
||||
input_ids = tokenizer.encode(prompt, return_tensors="pt").to('xpu')
|
||||
# ipex 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()
|
||||
output = model.generate(input_ids,
|
||||
max_new_tokens=args.n_predict)
|
||||
torch.xpu.synchronize()
|
||||
end = time.time()
|
||||
output_str = tokenizer.decode(output[0], skip_special_tokens=True)
|
||||
print(f'Inference time: {end-st} s')
|
||||
print('-'*20, 'Prompt', '-'*20)
|
||||
print(prompt)
|
||||
print('-'*20, 'Output', '-'*20)
|
||||
print(output_str)
|
||||
Loading…
Reference in a new issue