diff --git a/python/llm/example/GPU/PyTorch-Models/Model/README.md b/python/llm/example/GPU/PyTorch-Models/Model/README.md index 4dd7c311..6edf4077 100644 --- a/python/llm/example/GPU/PyTorch-Models/Model/README.md +++ b/python/llm/example/GPU/PyTorch-Models/Model/README.md @@ -5,6 +5,8 @@ You can use `optimize_model` API to accelerate general PyTorch models on Intel G | Model | Example | |----------------|----------------------------------------------------------| | Mistral | [link](mistral) | +| LLaMA 2 | [link](llama2) | +| ChatGLM2 | [link](chatglm2) | ## Verified Hardware Platforms diff --git a/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/README.md b/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/README.md new file mode 100644 index 00000000..be03f814 --- /dev/null +++ b/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/README.md @@ -0,0 +1,109 @@ +# ChatGLM2 +In this directory, you will find examples on how you could use BigDL-LLM `optimize_model` API to accelerate ChatGLM2 models. For illustration purposes, we utilize the [THUDM/chatglm2-6b](https://huggingface.co/THUDM/chatglm2-6b) as reference ChatGLM2 models. + +## 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 1: Predict Tokens using `generate()` API +In the example [generate.py](./generate.py), we show a basic use case for a ChatGLM2 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 +``` + +### 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 --prompt 'AI是什么？' +``` + +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 ChatGLM2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/chatglm2-6b'`. +- `--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`. + +#### 2.3 Sample Output +#### [THUDM/chatglm2-6b](https://huggingface.co/THUDM/chatglm2-6b) +```log +Inference time: xxxx s +-------------------- Output -------------------- +问：AI是什么？ + +答： AI指的是人工智能,是一种能够通过学习和推理来执行任务的计算机程序。AI可以分为弱人工智能和强人工智能。 + +弱人工智能(也称为狭 +``` + +```log +Inference time: xxxx s +-------------------- Output -------------------- +问：What is AI? + +答： Artificial Intelligence (AI) refers to the ability of a computer or machine to perform tasks that typically require human-like intelligence, such as understanding language, recognizing patterns +``` + +## Example 2: Stream Chat using `stream_chat()` API +In the example [streamchat.py](./streamchat.py), we show a basic use case for a ChatGLM2 model to stream chat, 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 + +# 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 +``` + +### 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 +``` + +**Stream Chat using `stream_chat()` API**: +``` +python ./streamchat.py +``` + +**Chat using `chat()` API**: +``` +python ./streamchat.py --disable-stream +``` + +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 ChatGLM2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/chatglm2-6b'`. +- `--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. diff --git a/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/generate.py b/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/generate.py new file mode 100644 index 00000000..8520f3b0 --- /dev/null +++ b/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/generate.py @@ -0,0 +1,74 @@ +# +# 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 AutoModel, AutoTokenizer +from bigdl.llm import optimize_model + +# you could tune the prompt based on your own model, +# here the prompt tuning refers to https://huggingface.co/THUDM/chatglm2-6b/blob/main/modeling_chatglm.py#L1007 +CHATGLM_V2_PROMPT_FORMAT = "问：{prompt}\n\n答：" + +if __name__ == '__main__': + parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for ChatGLM2 model') + parser.add_argument('--repo-id-or-model-path', type=str, default="THUDM/chatglm2-6b", + help='The huggingface repo id for the ChatGLM2 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 + model = AutoModel.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 = CHATGLM_V2_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) diff --git a/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/streamchat.py b/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/streamchat.py new file mode 100644 index 00000000..18ae779e --- /dev/null +++ b/python/llm/example/GPU/PyTorch-Models/Model/chatglm2/streamchat.py @@ -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 +import numpy as np + +from transformers import AutoModel, AutoTokenizer +from bigdl.llm import optimize_model + + +if __name__ == '__main__': + parser = argparse.ArgumentParser(description='Stream Chat for ChatGLM2 model') + parser.add_argument('--repo-id-or-model-path', type=str, default="THUDM/chatglm2-6b", + help='The huggingface repo id for the ChatGLM2 model to be downloaded' + ', or the path to the huggingface 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') + + args = parser.parse_args() + model_path = args.repo_id_or_model_path + disable_stream = args.disable_stream + + # Load model + model = AutoModel.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.to('xpu') + + # Load tokenizer + tokenizer = AutoTokenizer.from_pretrained(model_path, + trust_remote_code=True) + + with torch.inference_mode(): + prompt = args.question + 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=32) + + # start inference + if disable_stream: + # Chat + response, history = model.chat(tokenizer, args.question, history=[]) + print('-'*20, 'Chat Output', '-'*20) + print(response) + else: + # Stream chat + response_ = "" + print('-'*20, 'Stream Chat Output', '-'*20) + for response, history in model.stream_chat(tokenizer, args.question, history=[]): + print(response.replace(response_, ""), end="") + response_ = response diff --git a/python/llm/example/GPU/PyTorch-Models/Model/llama2/README.md b/python/llm/example/GPU/PyTorch-Models/Model/llama2/README.md new file mode 100644 index 00000000..f36d373d --- /dev/null +++ b/python/llm/example/GPU/PyTorch-Models/Model/llama2/README.md @@ -0,0 +1,69 @@ +# Llama2 +In this directory, you will find examples on how you could use BigDL-LLM `optimize_model` API to accelerate Llama2 models. For illustration purposes, we utilize the [meta-llama/Llama-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf) and [meta-llama/Llama-2-13b-chat-hf](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as reference Llama2 models. + +## 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 Llama2 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 +``` + +### 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 --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 Llama2 model (e.g. `meta-llama/Llama-2-7b-chat-hf` and `meta-llama/Llama-2-13b-chat-hf`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'meta-llama/Llama-2-7b-chat-hf'`. +- `--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 +#### [meta-llama/Llama-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf) +```log +Inference time: xxxx s +-------------------- Output -------------------- +### HUMAN: +What is AI? + +### RESPONSE: + +AI is a field of computer science that focuses on creating intelligent machines that can perform tasks that typically require human intelligence, such as understanding natural language, +``` + +#### [meta-llama/Llama-2-13b-chat-hf](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) +```log +Inference time: xxxx s +-------------------- Output -------------------- +### HUMAN: +What is AI? + +### RESPONSE: + +AI, or artificial intelligence, refers to the ability of machines to perform tasks that would typically require human intelligence, such as learning, problem-solving, +``` diff --git a/python/llm/example/GPU/PyTorch-Models/Model/llama2/generate.py b/python/llm/example/GPU/PyTorch-Models/Model/llama2/generate.py new file mode 100644 index 00000000..13ac5b50 --- /dev/null +++ b/python/llm/example/GPU/PyTorch-Models/Model/llama2/generate.py @@ -0,0 +1,78 @@ +# +# 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, +# here the prompt tuning refers to https://huggingface.co/georgesung/llama2_7b_chat_uncensored#prompt-style +LLAMA2_PROMPT_FORMAT = """### HUMAN: +{prompt} + +### RESPONSE: +""" + +if __name__ == '__main__': + parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for Llama2 model') + parser.add_argument('--repo-id-or-model-path', type=str, default="meta-llama/Llama-2-7b-chat-hf", + help='The huggingface repo id for the Llama2 (e.g. `meta-llama/Llama-2-7b-chat-hf` and `meta-llama/Llama-2-13b-chat-hf`) 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 = LLAMA2_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)