LLM: add gpu pytorch-models example llama2 and chatglm2 (#9142)

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
 

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.

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)

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

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,
+```

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)