Add cpu and gpu examples for SOLAR-10.7B (#9821)
This commit is contained in:
Jinyi Wan
GitHub
parent
ad6b182916
commit
3147ebe63d
9 changed files with 578 additions and 0 deletions
|
|
@ -74,6 +74,7 @@ Over 20 models have been optimized/verified on `bigdl-llm`, including *LLaMA/LLa
|
||||||
| Distil-Whisper | [link](example/CPU/HF-Transformers-AutoModels/Model/distil-whisper) | [link](example/GPU/HF-Transformers-AutoModels/Model/distil-whisper) |
|
| Distil-Whisper | [link](example/CPU/HF-Transformers-AutoModels/Model/distil-whisper) | [link](example/GPU/HF-Transformers-AutoModels/Model/distil-whisper) |
|
||||||
| Yi | [link](example/CPU/HF-Transformers-AutoModels/Model/yi) | [link](example/GPU/HF-Transformers-AutoModels/Model/yi) |
|
| Yi | [link](example/CPU/HF-Transformers-AutoModels/Model/yi) | [link](example/GPU/HF-Transformers-AutoModels/Model/yi) |
|
||||||
| BlueLM | [link](example/CPU/HF-Transformers-AutoModels/Model/bluelm) | [link](example/GPU/HF-Transformers-AutoModels/Model/bluelm) |
|
| BlueLM | [link](example/CPU/HF-Transformers-AutoModels/Model/bluelm) | [link](example/GPU/HF-Transformers-AutoModels/Model/bluelm) |
|
||||||
|
| Solar-10.7B | [link](example/CPU/HF-Transformers-AutoModels/Model/solar-10.7b) | [link](example/GPU/HF-Transformers-AutoModels/Model/solar-10.7b) |
|
||||||
|
|
||||||
### Working with `bigdl-llm`
|
### Working with `bigdl-llm`
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,80 @@
|
||||||
|
# SOLAR-10.7B
|
||||||
|
In this directory, you will find examples on how you could apply BigDL-LLM INT4 optimizations on SOLAR-10.7B models. For illustration purposes, we utilize the [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0) as a reference SOLAR-10.7B model.
|
||||||
|
|
||||||
|
## 0. Requirements
|
||||||
|
To run these examples with BigDL-LLM, we have some recommended requirements for your machine, please refer to [here](../README.md#recommended-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 SOLAR-10.7B model to predict the next N tokens using `generate()` API, with BigDL-LLM INT4 optimizations.
|
||||||
|
### 1. Install
|
||||||
|
We suggest using conda to manage environment:
|
||||||
|
```bash
|
||||||
|
conda create -n llm python=3.9
|
||||||
|
conda activate llm
|
||||||
|
|
||||||
|
pip install --pre --upgrade bigdl-llm[all] # install the latest bigdl-llm nightly build with 'all' option
|
||||||
|
pip install transformers==4.35.2 # required by SOLAR-10.7B
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. Run
|
||||||
|
```
|
||||||
|
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 SOLAR-10.7B model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'upstage/SOLAR-10.7B-Instruct-v1.0'`.
|
||||||
|
- `--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`.
|
||||||
|
|
||||||
|
> **Note**: When loading the model in 4-bit, BigDL-LLM converts linear layers in the model into INT4 format. In theory, a *X*B model saved in 16-bit will requires approximately 2*X* GB of memory for loading, and ~0.5*X* GB memory for further inference.
|
||||||
|
>
|
||||||
|
> Please select the appropriate size of the SOLAR-10.7B model based on the capabilities of your machine.
|
||||||
|
|
||||||
|
#### 2.1 Client
|
||||||
|
On client Windows machine, it is recommended to run directly with full utilization of all cores:
|
||||||
|
```powershell
|
||||||
|
python ./generate.py
|
||||||
|
```
|
||||||
|
|
||||||
|
#### 2.2 Server
|
||||||
|
For optimal performance on server, it is recommended to set several environment variables (refer to [here](../README.md#best-known-configuration-on-linux) for more information), and run the example with all the physical cores of a single socket.
|
||||||
|
|
||||||
|
E.g. on Linux,
|
||||||
|
```bash
|
||||||
|
# set BigDL-LLM env variables
|
||||||
|
source bigdl-llm-init
|
||||||
|
|
||||||
|
# e.g. for a server with 48 cores per socket
|
||||||
|
export OMP_NUM_THREADS=48
|
||||||
|
numactl -C 0-47 -m 0 python ./generate.py
|
||||||
|
```
|
||||||
|
|
||||||
|
#### 2.3 Sample Output
|
||||||
|
#### [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0)
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Prompt --------------------
|
||||||
|
<s>### User:
|
||||||
|
What is AI?
|
||||||
|
### Assistant:
|
||||||
|
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
What is AI?
|
||||||
|
### Assistant:
|
||||||
|
AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It involves the development of
|
||||||
|
```
|
||||||
|
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Prompt --------------------
|
||||||
|
<s>### User:
|
||||||
|
AI是什么？
|
||||||
|
### Assistant:
|
||||||
|
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
AI是什么？
|
||||||
|
### Assistant:
|
||||||
|
AI, 全称为人工智能(Artificial Intelligence)，是计算机科学、心理学、语言学、逻
|
||||||
|
```
|
||||||
|
|
@ -0,0 +1,71 @@
|
||||||
|
#
|
||||||
|
# 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,
|
||||||
|
# prompt format is tuned based on the output example in this link:
|
||||||
|
# https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0#usage-instructions
|
||||||
|
SOLAR_PROMPT_FORMAT = "<s>### User:\n{prompt}\n### Assistant:\n"
|
||||||
|
|
||||||
|
if __name__ == '__main__':
|
||||||
|
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for SOLAR-10.7B model')
|
||||||
|
parser.add_argument('--repo-id-or-model-path', type=str, default="upstage/SOLAR-10.7B-Instruct-v1.0",
|
||||||
|
help='The huggingface repo id for the SOLAR-10.7B model 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
|
||||||
|
model = AutoModelForCausalLM.from_pretrained(model_path,
|
||||||
|
load_in_4bit=True,
|
||||||
|
trust_remote_code=True)
|
||||||
|
|
||||||
|
# Load tokenizer
|
||||||
|
tokenizer = AutoTokenizer.from_pretrained(model_path,
|
||||||
|
trust_remote_code=True)
|
||||||
|
|
||||||
|
# Generate predicted tokens
|
||||||
|
with torch.inference_mode():
|
||||||
|
prompt = SOLAR_PROMPT_FORMAT.format(prompt=args.prompt)
|
||||||
|
input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
|
||||||
|
|
||||||
|
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 BigDL-LLM INT4 optimizations
|
||||||
|
output = model.generate(input_ids,
|
||||||
|
max_new_tokens=args.n_predict)
|
||||||
|
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)
|
||||||
|
|
@ -0,0 +1,70 @@
|
||||||
|
# SOLAR-10.7B
|
||||||
|
In this directory, you will find examples on how you could use BigDL-LLM `optimize_model` API to accelerate SOLAR-10.7B models. For illustration purposes, we utilize the [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0) as a reference SOLAR-10.7B model.
|
||||||
|
|
||||||
|
## Requirements
|
||||||
|
To run these examples with BigDL-LLM, we have some recommended requirements for your machine, please refer to [here](../README.md#recommended-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 SOLAR-10.7B model to predict the next N tokens using `generate()` API, with BigDL-LLM INT4 optimizations.
|
||||||
|
### 1. Install
|
||||||
|
We suggest using conda to manage the Python environment. For more information about conda installation, please refer to [here](https://docs.conda.io/en/latest/miniconda.html#).
|
||||||
|
|
||||||
|
After installing conda, create a Python environment for BigDL-LLM:
|
||||||
|
```bash
|
||||||
|
conda create -n llm python=3.9 # recommend to use Python 3.9
|
||||||
|
conda activate llm
|
||||||
|
|
||||||
|
pip install --pre --upgrade bigdl-llm[all] # install the latest bigdl-llm nightly build with 'all' option
|
||||||
|
pip install transformers==4.35.2 # required by SOLAR-10.7B
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. Run
|
||||||
|
After setting up the Python environment, you could run the example by following steps.
|
||||||
|
|
||||||
|
#### 2.1 Client
|
||||||
|
On client Windows machines, it is recommended to run directly with full utilization of all cores:
|
||||||
|
```powershell
|
||||||
|
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
|
||||||
|
```
|
||||||
|
More information about arguments can be found in [Arguments Info](#23-arguments-info) section. The expected output can be found in [Sample Output](#24-sample-output) section.
|
||||||
|
|
||||||
|
#### 2.2 Server
|
||||||
|
For optimal performance on server, it is recommended to set several environment variables (refer to [here](../README.md#best-known-configuration-on-linux) for more information), and run the example with all the physical cores of a single socket.
|
||||||
|
|
||||||
|
E.g. on Linux,
|
||||||
|
```bash
|
||||||
|
# set BigDL-LLM env variables
|
||||||
|
source bigdl-llm-init
|
||||||
|
|
||||||
|
# e.g. for a server with 48 cores per socket
|
||||||
|
export OMP_NUM_THREADS=48
|
||||||
|
numactl -C 0-47 -m 0 python ./generate.py --prompt 'What is AI?'
|
||||||
|
```
|
||||||
|
More information about arguments can be found in [Arguments Info](#23-arguments-info) section. The expected output can be found in [Sample Output](#24-sample-output) section.
|
||||||
|
|
||||||
|
#### 2.3 Arguments Info
|
||||||
|
In the example, several arguments can be passed to satisfy your requirements:
|
||||||
|
|
||||||
|
- `--repo-id-or-model-path`: str, argument defining the huggingface repo id for the SOLAR-10.7B model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'upstage/SOLAR-10.7B-Instruct-v1.0'`.
|
||||||
|
- `--prompt`: str, argument defining the prompt to be inferred (with integrated prompt format for chat). It is default to be `'What is AI?'`.
|
||||||
|
- `--n-predict`: int, argument defining the max number of tokens to predict. It is default to be `32`.
|
||||||
|
|
||||||
|
#### 2.4 Sample Output
|
||||||
|
#### [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0)
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
What is AI?
|
||||||
|
### Assistant:
|
||||||
|
AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It involves the development of
|
||||||
|
```
|
||||||
|
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
AI是什么？
|
||||||
|
### Assistant:
|
||||||
|
AI, 全称为人工智能(Artificial Intelligence)，是计算机科学、心理学、语言学、逻
|
||||||
|
```
|
||||||
|
|
@ -0,0 +1,65 @@
|
||||||
|
#
|
||||||
|
# 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 transformers import AutoModelForCausalLM, AutoTokenizer
|
||||||
|
from bigdl.llm import optimize_model
|
||||||
|
|
||||||
|
# you could tune the prompt based on your own model,
|
||||||
|
# prompt format is tuned based on the output example in this link:
|
||||||
|
# https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0#usage-instructions
|
||||||
|
SOLAR_PROMPT_FORMAT = "<s>### User:\n{prompt}\n### Assistant:\n"
|
||||||
|
|
||||||
|
if __name__ == '__main__':
|
||||||
|
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for SOLAR-10.7B model')
|
||||||
|
parser.add_argument('--repo-id-or-model-path', type=str, default="upstage/SOLAR-10.7B-Instruct-v1.0",
|
||||||
|
help='The huggingface repo id for the SOLAR-10.7B model 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
|
||||||
|
model = AutoModelForCausalLM.from_pretrained(model_path,
|
||||||
|
device_map="cpu",
|
||||||
|
torch_dtype=torch.float16,
|
||||||
|
trust_remote_code=True)
|
||||||
|
|
||||||
|
# With only one line to enable BigDL-LLM optimization on model
|
||||||
|
model = optimize_model(model)
|
||||||
|
|
||||||
|
# Load tokenizer
|
||||||
|
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
|
||||||
|
|
||||||
|
# Generate predicted tokens
|
||||||
|
with torch.inference_mode():
|
||||||
|
prompt = SOLAR_PROMPT_FORMAT.format(prompt=args.prompt)
|
||||||
|
input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
|
||||||
|
st = time.time()
|
||||||
|
output = model.generate(input_ids,
|
||||||
|
max_new_tokens=args.n_predict)
|
||||||
|
end = time.time()
|
||||||
|
output_str = tokenizer.decode(output[0], skip_special_tokens=True)
|
||||||
|
print(f'Inference time: {end-st} s')
|
||||||
|
print('-'*20, 'Output', '-'*20)
|
||||||
|
print(output_str)
|
||||||
|
|
@ -0,0 +1,72 @@
|
||||||
|
# SOLAR-10.7B
|
||||||
|
In this directory, you will find examples on how you could apply BigDL-LLM INT4 optimizations on SOLAR-10.7B models on [Intel GPUs](../README.md). For illustration purposes, we utilize the [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0) as a reference SOLAR-10.7B 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#recommended-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 SOLAR-10.7B model to predict the next N tokens using `generate()` API, with BigDL-LLM INT4 optimizations on Intel GPUs.
|
||||||
|
### 1. Install
|
||||||
|
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==2.0.110+xpu as default
|
||||||
|
# you can install specific ipex/torch version for your need
|
||||||
|
pip install --pre --upgrade bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu
|
||||||
|
pip install transformers==4.35.2 # required by SOLAR-10.7B
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. Configures OneAPI environment variables
|
||||||
|
```bash
|
||||||
|
source /opt/intel/oneapi/setvars.sh
|
||||||
|
```
|
||||||
|
|
||||||
|
### 3. Run
|
||||||
|
|
||||||
|
For optimal performance on Arc, it is recommended to set several environment variables.
|
||||||
|
|
||||||
|
```bash
|
||||||
|
export USE_XETLA=OFF
|
||||||
|
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
|
||||||
|
```
|
||||||
|
|
||||||
|
```bash
|
||||||
|
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 SOLAR-10.7B model (e.g `upstage/SOLAR-10.7B-Instruct-v1.0`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'upstage/SOLAR-10.7B-Instruct-v1.0'`.
|
||||||
|
- `--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
|
||||||
|
#### [upstage/SOLAR-10.7B-Instruct-v1.0t](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0)
|
||||||
|
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Prompt --------------------
|
||||||
|
<s>### User:
|
||||||
|
What is AI?
|
||||||
|
### Assistant:
|
||||||
|
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
What is AI?
|
||||||
|
### Assistant:
|
||||||
|
AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It involves the development of
|
||||||
|
```
|
||||||
|
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Prompt --------------------
|
||||||
|
<s>### User:
|
||||||
|
AI是什么？
|
||||||
|
### Assistant:
|
||||||
|
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
AI是什么？
|
||||||
|
### Assistant:
|
||||||
|
AI, 全称为人工智能(Artificial Intelligence)，是计算机科学、心理学、语言学、逻
|
||||||
|
```
|
||||||
|
|
@ -0,0 +1,79 @@
|
||||||
|
#
|
||||||
|
# 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 intel_extension_for_pytorch as ipex
|
||||||
|
import time
|
||||||
|
import argparse
|
||||||
|
|
||||||
|
from bigdl.llm.transformers import AutoModelForCausalLM
|
||||||
|
from transformers import AutoTokenizer
|
||||||
|
|
||||||
|
# you could tune the prompt based on your own model,
|
||||||
|
# prompt format is tuned based on the output example in this link:
|
||||||
|
# https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0#usage-instructions
|
||||||
|
SOLAR_PROMPT_FORMAT = "<s>### User:\n{prompt}\n### Assistant:\n"
|
||||||
|
|
||||||
|
if __name__ == '__main__':
|
||||||
|
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for SOLAR-10.7B model')
|
||||||
|
parser.add_argument('--repo-id-or-model-path', type=str, default="upstage/SOLAR-10.7B-Instruct-v1.0",
|
||||||
|
help='The huggingface repo id for the SOLAR-10.7B model 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
|
||||||
|
model = AutoModelForCausalLM.from_pretrained(model_path,
|
||||||
|
load_in_4bit=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 = SOLAR_PROMPT_FORMAT.format(prompt=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()
|
||||||
|
# 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 BigDL-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, 'Prompt', '-'*20)
|
||||||
|
print(prompt)
|
||||||
|
print('-'*20, 'Output', '-'*20)
|
||||||
|
print(output_str)
|
||||||
|
|
@ -0,0 +1,65 @@
|
||||||
|
# SOLAR-10.7B
|
||||||
|
In this directory, you will find examples on how you could use BigDL-LLM `optimize_model` API to accelerate SOLAR-10.7B models. For illustration purposes, we utilize the [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0) as a reference SOLAR-10.7B model.
|
||||||
|
|
||||||
|
## 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#recommended-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 SOLAR-10.7B model to predict the next N tokens using `generate()` API, with BigDL-LLM INT4 optimizations on Intel GPUs.
|
||||||
|
### 1. Install
|
||||||
|
We suggest using conda to manage the Python environment. For more information about conda installation, please refer to [here](https://docs.conda.io/en/latest/miniconda.html#).
|
||||||
|
|
||||||
|
After installing conda, create a Python environment for BigDL-LLM:
|
||||||
|
```bash
|
||||||
|
conda create -n llm python=3.9 # recommend to use Python 3.9
|
||||||
|
conda activate llm
|
||||||
|
|
||||||
|
# below command will install intel_extension_for_pytorch==2.0.110+xpu as default
|
||||||
|
# you can install specific ipex/torch version for your need
|
||||||
|
pip install --pre --upgrade bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu
|
||||||
|
pip install transformers==4.35.2 # required by SOLAR-10.7B
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. Configures OneAPI environment variables
|
||||||
|
```bash
|
||||||
|
source /opt/intel/oneapi/setvars.sh
|
||||||
|
```
|
||||||
|
|
||||||
|
### 3. Run
|
||||||
|
|
||||||
|
For optimal performance on Arc, it is recommended to set several environment variables.
|
||||||
|
|
||||||
|
```bash
|
||||||
|
export USE_XETLA=OFF
|
||||||
|
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
|
||||||
|
```
|
||||||
|
|
||||||
|
```bash
|
||||||
|
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
|
||||||
|
```
|
||||||
|
|
||||||
|
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 SOLAR-10.7B model (e.g `upstage/SOLAR-10.7B-Instruct-v1.0`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'upstage/SOLAR-10.7B-Instruct-v1.0'`.
|
||||||
|
- `--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`.
|
||||||
|
|
||||||
|
#### 2.3 Sample Output
|
||||||
|
#### [upstage/SOLAR-10.7B-Instruct-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0)
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
What is AI?
|
||||||
|
### Assistant:
|
||||||
|
AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It involves the development of
|
||||||
|
```
|
||||||
|
|
||||||
|
```log
|
||||||
|
Inference time: XXXX s
|
||||||
|
-------------------- Output --------------------
|
||||||
|
### User:
|
||||||
|
AI是什么？
|
||||||
|
### Assistant:
|
||||||
|
AI, 全称为人工智能(Artificial Intelligence)，是计算机科学、心理学、语言学、逻
|
||||||
|
```
|
||||||
|
|
@ -0,0 +1,75 @@
|
||||||
|
#
|
||||||
|
# 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 intel_extension_for_pytorch as ipex
|
||||||
|
import time
|
||||||
|
import argparse
|
||||||
|
|
||||||
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
||||||
|
from bigdl.llm import optimize_model
|
||||||
|
|
||||||
|
# you could tune the prompt based on your own model,
|
||||||
|
# prompt format is tuned based on the output example in this link:
|
||||||
|
# https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0#usage-instructions
|
||||||
|
SOLAR_PROMPT_FORMAT = "<s>### User:\n{prompt}\n### Assistant:\n"
|
||||||
|
|
||||||
|
if __name__ == '__main__':
|
||||||
|
parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for SOLAR-10.7B model')
|
||||||
|
parser.add_argument('--repo-id-or-model-path', type=str, default="upstage/SOLAR-10.7B-Instruct-v1.0",
|
||||||
|
help='The huggingface repo id for the SOLAR-10.7B model 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
|
||||||
|
model = AutoModelForCausalLM.from_pretrained(model_path,
|
||||||
|
trust_remote_code=True,
|
||||||
|
torch_dtype='auto',
|
||||||
|
low_cpu_mem_usage=True)
|
||||||
|
|
||||||
|
# With only one line to enable BigDL-LLM optimization on model
|
||||||
|
model = optimize_model(model)
|
||||||
|
model = model.to('xpu')
|
||||||
|
|
||||||
|
# Load tokenizer
|
||||||
|
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
|
||||||
|
|
||||||
|
# Generate predicted tokens
|
||||||
|
with torch.inference_mode():
|
||||||
|
prompt = SOLAR_PROMPT_FORMAT.format(prompt=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 = 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