| Qwen-VL |
diff --git a/python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen2/README.md b/python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen2/README.md
new file mode 100644
index 00000000..423de9b4
--- /dev/null
+++ b/python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen2/README.md
@@ -0,0 +1,83 @@
+# Qwen2
+
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Qwen2 models. For illustration purposes, we utilize the [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct) as a reference Qwen2 model.
+
+## 0. Requirements
+To run these examples with IPEX-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 Qwen model to predict the next N tokens using `generate()` API, with IPEX-LLM INT4 optimizations.
+### 1. Install
+We suggest using conda to manage environment:
+
+On Linux:
+
+```bash
+conda create -n llm python=3.11
+conda activate llm
+
+# install ipex-llm with 'all' option
+pip install --pre --upgrade ipex-llm[all] --extra-index-url https://download.pytorch.org/whl/cpu
+pip install transformers==4.37.0 # install the transformers which support Qwen2
+```
+
+On Windows:
+
+```cmd
+conda create -n llm python=3.11
+conda activate llm
+
+pip install --pre --upgrade ipex-llm[all]
+pip install transformers==4.37.0
+```
+
+### 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 Qwen model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'Qwen/Qwen2-7B-Instruct'`.
+- `--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`.
+
+> **Note**: When loading the model in 4-bit, IPEX-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 Qwen 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:
+```cmd
+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 IPEX-LLM env variables
+source ipex-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
+##### [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+AI是什么？
+-------------------- Output --------------------
+AI，即人工智能（Artificial Intelligence），是一种计算机科学领域，旨在开发能够模拟、延伸和增强人类智能的算法和系统。人工 智能涉及许多
+```
+
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+What is AI?
+-------------------- Output --------------------
+AI stands for Artificial Intelligence, which is the simulation of human intelligence in machines that are programmed to think and learn like humans and mimic their actions. The term may
+```
\ No newline at end of file
diff --git a/python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen2/generate.py b/python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen2/generate.py
new file mode 100644
index 00000000..90626539
--- /dev/null
+++ b/python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen2/generate.py
@@ -0,0 +1,80 @@
+#
+# 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 transformers import AutoTokenizer
+
+
+if __name__ == '__main__':
+ parser = argparse.ArgumentParser(description='Qwen2-7B-Instruct')
+ parser.add_argument('--repo-id-or-model-path', type=str, default="Qwen/Qwen2-7B-Instruct",
+ help='The huggingface repo id for the Qwen2 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
+
+
+ from ipex_llm.transformers import AutoModelForCausalLM
+ model = AutoModelForCausalLM.from_pretrained(model_path,
+ load_in_4bit=True,
+ optimize_model=True,
+ trust_remote_code=True,
+ use_cache=True)
+
+ # Load tokenizer
+ tokenizer = AutoTokenizer.from_pretrained(model_path,
+ trust_remote_code=True)
+
+ prompt = args.prompt
+
+ # Generate predicted tokens
+ with torch.inference_mode():
+ # The following code for generation is adapted from https://huggingface.co/Qwen/Qwen2-7B-Instruct#quickstart
+ messages = [
+ {"role": "system", "content": "You are a helpful assistant."},
+ {"role": "user", "content": prompt}
+ ]
+ text = tokenizer.apply_chat_template(
+ messages,
+ tokenize=False,
+ add_generation_prompt=True
+ )
+ model_inputs = tokenizer([text], return_tensors="pt")
+ st = time.time()
+ generated_ids = model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=args.n_predict
+ )
+ end = time.time()
+ generated_ids = [
+ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+ ]
+
+ response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+ print(f'Inference time: {end-st} s')
+ print('-'*20, 'Prompt', '-'*20)
+ print(prompt)
+ print('-'*20, 'Output', '-'*20)
+ print(response)
diff --git a/python/llm/example/CPU/PyTorch-Models/Model/qwen2/README.md b/python/llm/example/CPU/PyTorch-Models/Model/qwen2/README.md
new file mode 100644
index 00000000..84d7cc12
--- /dev/null
+++ b/python/llm/example/CPU/PyTorch-Models/Model/qwen2/README.md
@@ -0,0 +1,84 @@
+# Qwen2
+In this directory, you will find examples on how you could use IPEX-LLM `optimize_model` API to accelerate Qwen2 models. For illustration purposes, we utilize the [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct) as reference Qwen2 model.
+
+## Requirements
+To run these examples with IPEX-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 Qwen2 model to predict the next N tokens using `generate()` API, with IPEX-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://conda-forge.org/download/).
+
+After installing conda, create a Python environment for IPEX-LLM:
+
+On Linux:
+
+```bash
+conda create -n llm python=3.11 # recommend to use Python 3.11
+conda activate llm
+
+# install the latest ipex-llm nightly build with 'all' option
+pip install --pre --upgrade ipex-llm[all] --extra-index-url https://download.pytorch.org/whl/cpu
+pip install transformers==4.37.0 # install transformers which supports Qwen2
+```
+
+On Windows:
+
+```cmd
+conda create -n llm python=3.11
+conda activate llm
+
+pip install --pre --upgrade ipex-llm[all]
+pip install transformers==4.37.0
+```
+
+### 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 Qwen2 to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'Qwen/Qwen2-7B-Instruct'`.
+- `--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`.
+
+> **Note**: When loading the model in 4-bit, IPEX-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 Qwen model based on the capabilities of your machine.
+
+#### 2.1 Client
+On client Windows machines, it is recommended to run directly with full utilization of all cores:
+```cmd
+python ./generate.py --prompt 'What is AI?'
+```
+
+#### 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 IPEX-LLM env variables
+source ipex-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?'
+```
+
+#### 2.3 Sample Output
+##### [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+AI是什么？
+-------------------- Output --------------------
+AI，即人工智能（Artificial Intelligence），是一门研究、开发用于模拟、延伸和扩展人的智能的理论、方法、技术及应用系统的学科
+```
+
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+What is AI?
+-------------------- Output --------------------
+AI stands for Artificial Intelligence, which refers to the development of computer systems that can perform tasks that typically require human intelligence. These tasks may include learning from experience,
+```
diff --git a/python/llm/example/CPU/PyTorch-Models/Model/qwen2/generate.py b/python/llm/example/CPU/PyTorch-Models/Model/qwen2/generate.py
new file mode 100644
index 00000000..0c9b428a
--- /dev/null
+++ b/python/llm/example/CPU/PyTorch-Models/Model/qwen2/generate.py
@@ -0,0 +1,82 @@
+#
+# 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 transformers import AutoTokenizer
+
+
+if __name__ == '__main__':
+ parser = argparse.ArgumentParser(description='Qwen2-7B-Instruct')
+ parser.add_argument('--repo-id-or-model-path', type=str, default="Qwen/Qwen2-7B-Instruct",
+ help='The huggingface repo id for the Qwen2 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
+
+
+ from transformers import AutoModelForCausalLM
+ model = AutoModelForCausalLM.from_pretrained(model_path,
+ trust_remote_code=True,
+ torch_dtype='auto',
+ low_cpu_mem_usage=True,
+ use_cache=True)
+
+
+ # Load tokenizer
+ tokenizer = AutoTokenizer.from_pretrained(model_path,
+ trust_remote_code=True)
+ from ipex_llm import optimize_model
+ model = optimize_model(model)
+
+ prompt = args.prompt
+ # Generate predicted tokens
+ with torch.inference_mode():
+ # The following code for generation is adapted from https://huggingface.co/Qwen/Qwen2-7B-Instruct#quickstart
+ messages = [
+ {"role": "system", "content": "You are a helpful assistant."},
+ {"role": "user", "content": prompt}
+ ]
+ text = tokenizer.apply_chat_template(
+ messages,
+ tokenize=False,
+ add_generation_prompt=True
+ )
+ model_inputs = tokenizer([text], return_tensors="pt")
+ st = time.time()
+ generated_ids = model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=args.n_predict
+ )
+ end = time.time()
+ generated_ids = [
+ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+ ]
+
+ response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+ print(f'Inference time: {end-st} s')
+ print('-'*20, 'Prompt', '-'*20)
+ print(prompt)
+ print('-'*20, 'Output', '-'*20)
+ print(response)
diff --git a/python/llm/example/GPU/HF-Transformers-AutoModels/Model/qwen2/README.md b/python/llm/example/GPU/HF-Transformers-AutoModels/Model/qwen2/README.md
new file mode 100644
index 00000000..274b0b47
--- /dev/null
+++ b/python/llm/example/GPU/HF-Transformers-AutoModels/Model/qwen2/README.md
@@ -0,0 +1,134 @@
+# Qwen2
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Qwen2 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct) as a reference InternLM 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.
+
+## Example: Predict Tokens using `generate()` API
+In the example [generate.py](./generate.py), we show a basic use case for a Qwen2 model to predict the next N tokens using `generate()` API, with IPEX-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=3.11
+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/
+
+pip install transformers==4.37.0 # install transformers which supports Qwen2
+```
+
+#### 1.2 Installation on Windows
+We suggest using conda to manage environment:
+```bash
+conda create -n llm python=3.11 libuv
+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/
+
+pip install transformers==4.37.0 # install transformers which supports Qwen2
+```
+
+### 2. Configures OneAPI environment variables for Linux
+
+> [!NOTE]
+> Skip this step if you are running on Windows.
+
+This is a required step on Linux for APT or offline installed oneAPI. Skip this step for PIP-installed oneAPI.
+
+```bash
+source /opt/intel/oneapi/setvars.sh
+```
+
+### 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
+
+
+For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series
+
+```bash
+export USE_XETLA=OFF
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export SYCL_CACHE_PERSISTENT=1
+```
+
+
+
+
+
+For Intel Data Center GPU Max Series
+
+```bash
+export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export SYCL_CACHE_PERSISTENT=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`.
+
+
+
+
+For Intel iGPU
+
+```bash
+export SYCL_CACHE_PERSISTENT=1
+export BIGDL_LLM_XMX_DISABLED=1
+```
+
+
+
+#### 3.2 Configurations for Windows
+
+
+For Intel iGPU
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+set BIGDL_LLM_XMX_DISABLED=1
+```
+
+
+
+
+
+For Intel Arc™ A-Series Graphics
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+```
+
+
+
+> [!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 Qwen2 model (e.g. `Qwen/Qwen2-7B-Instruct`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'Qwen/Qwen2-7B-Instruct'`.
+- `--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
+##### [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+AI是什么？
+-------------------- Output --------------------
+AI，即人工智能（Artificial Intelligence），是一种模拟人类智能行为的技术。它使计算机系统能够执行类似于人类的思维活动，例 如学习、推理
+```
+
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+What is AI?
+-------------------- Output --------------------
+AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans and mimic their actions. The term may
+```
\ No newline at end of file
diff --git a/python/llm/example/GPU/HF-Transformers-AutoModels/Model/qwen2/generate.py b/python/llm/example/GPU/HF-Transformers-AutoModels/Model/qwen2/generate.py
new file mode 100644
index 00000000..25fdaeec
--- /dev/null
+++ b/python/llm/example/GPU/HF-Transformers-AutoModels/Model/qwen2/generate.py
@@ -0,0 +1,92 @@
+#
+# 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 AutoTokenizer
+from ipex_llm import optimize_model
+import numpy as np
+
+
+if __name__ == '__main__':
+ parser = argparse.ArgumentParser(description='Qwen2-7B-Instruct')
+ parser.add_argument('--repo-id-or-model-path', type=str, default="Qwen/Qwen2-7B-Instruct",
+ help='The huggingface repo id for the Qwen2 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
+
+
+ from ipex_llm.transformers import AutoModelForCausalLM
+ # 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,
+ 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)
+
+ prompt = args.prompt
+
+ # Generate predicted tokens
+ with torch.inference_mode():
+ # The following code for generation is adapted from https://huggingface.co/Qwen/Qwen2-7B-Instruct#quickstart
+ messages = [
+ {"role": "system", "content": "You are a helpful assistant."},
+ {"role": "user", "content": prompt}
+ ]
+ text = tokenizer.apply_chat_template(
+ messages,
+ tokenize=False,
+ add_generation_prompt=True
+ )
+ model_inputs = tokenizer([text], return_tensors="pt").to("xpu")
+ # warmup
+ generated_ids = model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=args.n_predict
+ )
+
+ st = time.time()
+ generated_ids = model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=args.n_predict
+ )
+ torch.xpu.synchronize()
+ end = time.time()
+ generated_ids = generated_ids.cpu()
+ generated_ids = [
+ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+ ]
+
+ response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+ print(f'Inference time: {end-st} s')
+ print('-'*20, 'Prompt', '-'*20)
+ print(prompt)
+ print('-'*20, 'Output', '-'*20)
+ print(response)
\ No newline at end of file
diff --git a/python/llm/example/GPU/PyTorch-Models/Model/qwen2/README.md b/python/llm/example/GPU/PyTorch-Models/Model/qwen2/README.md
new file mode 100644
index 00000000..9a3e3e03
--- /dev/null
+++ b/python/llm/example/GPU/PyTorch-Models/Model/qwen2/README.md
@@ -0,0 +1,134 @@
+# Qwen2
+In this directory, you will find examples on how you could use IPEX-LLM `optimize_model` API to accelerate Qwen2 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct) as a reference InternLM 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.
+
+## Example: Predict Tokens using `generate()` API
+In the example [generate.py](./generate.py), we show a basic use case for a Qwen2 model to predict the next N tokens using `generate()` API, with IPEX-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=3.11
+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/
+
+pip install transformers==4.37.0 # install transformers which supports Qwen2
+```
+
+#### 1.2 Installation on Windows
+We suggest using conda to manage environment:
+```bash
+conda create -n llm python=3.11 libuv
+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/
+
+pip install transformers==4.37.0 # install transformers which supports Qwen2
+```
+
+### 2. Configures OneAPI environment variables for Linux
+
+> [!NOTE]
+> Skip this step if you are running on Windows.
+
+This is a required step on Linux for APT or offline installed oneAPI. Skip this step for PIP-installed oneAPI.
+
+```bash
+source /opt/intel/oneapi/setvars.sh
+```
+
+### 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
+
+
+For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series
+
+```bash
+export USE_XETLA=OFF
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export SYCL_CACHE_PERSISTENT=1
+```
+
+
+
+
+
+For Intel Data Center GPU Max Series
+
+```bash
+export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export SYCL_CACHE_PERSISTENT=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`.
+
+
+
+
+For Intel iGPU
+
+```bash
+export SYCL_CACHE_PERSISTENT=1
+export BIGDL_LLM_XMX_DISABLED=1
+```
+
+
+
+#### 3.2 Configurations for Windows
+
+
+For Intel iGPU
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+set BIGDL_LLM_XMX_DISABLED=1
+```
+
+
+
+
+
+For Intel Arc™ A-Series Graphics
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+```
+
+
+
+> [!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 Qwen2 model (e.g. `Qwen/Qwen2-7B-Instruct`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'Qwen/Qwen2-7B-Instruct'`.
+- `--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
+##### [Qwen/Qwen2-7B-Instruct](https://huggingface.co/Qwen/Qwen2-7B-Instruct)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+AI是什么？
+-------------------- Output --------------------
+AI是人工智能（Artificial Intelligence）的缩写。它指的是通过计算机程序、算法和模型来模拟、延伸和扩展人类智能的一门学科
+```
+
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+What is AI?
+-------------------- Output --------------------
+AI stands for Artificial Intelligence. It refers to the simulation of human intelligence in machines that are programmed to think and work like humans. This includes learning from experience,
+```
\ No newline at end of file
diff --git a/python/llm/example/GPU/PyTorch-Models/Model/qwen2/generate.py b/python/llm/example/GPU/PyTorch-Models/Model/qwen2/generate.py
new file mode 100644
index 00000000..c3c19253
--- /dev/null
+++ b/python/llm/example/GPU/PyTorch-Models/Model/qwen2/generate.py
@@ -0,0 +1,91 @@
+#
+# 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 AutoTokenizer
+from ipex_llm import optimize_model
+import numpy as np
+
+
+if __name__ == '__main__':
+ parser = argparse.ArgumentParser(description='Qwen2-7B-Instruct')
+ parser.add_argument('--repo-id-or-model-path', type=str, default="Qwen/Qwen2-7B-Instruct",
+ help='The huggingface repo id for the Qwen2 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
+
+
+ from transformers import AutoModelForCausalLM
+ from ipex_llm import optimize_model
+ model = AutoModelForCausalLM.from_pretrained(model_path,
+ trust_remote_code=True,
+ torch_dtype = 'auto',
+ low_cpu_mem_usage=True,
+ use_cache=True)
+ model = optimize_model(model)
+ model = model.to("xpu")
+
+ # Load tokenizer
+ tokenizer = AutoTokenizer.from_pretrained(model_path,
+ trust_remote_code=True)
+
+ prompt = args.prompt
+ # Generate predicted tokens
+ with torch.inference_mode():
+ # The following code for generation is adapted from https://huggingface.co/Qwen/Qwen2-7B-Instruct#quickstart
+ messages = [
+ {"role": "system", "content": "You are a helpful assistant."},
+ {"role": "user", "content": prompt}
+ ]
+ text = tokenizer.apply_chat_template(
+ messages,
+ tokenize=False,
+ add_generation_prompt=True
+ )
+ model_inputs = tokenizer([text], return_tensors="pt").to("xpu")
+ # warmup
+ generated_ids = model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=args.n_predict
+ )
+
+ st = time.time()
+ generated_ids = model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=args.n_predict
+ )
+ torch.xpu.synchronize()
+ end = time.time()
+ generated_ids = generated_ids.cpu()
+ generated_ids = [
+ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+ ]
+
+ response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+ print(f'Inference time: {end-st} s')
+ print('-'*20, 'Prompt', '-'*20)
+ print(prompt)
+ print('-'*20, 'Output', '-'*20)
+ print(response)
\ No newline at end of file
|