LLM: Add codegeex2 example (#11143)

* add codegeex example

* update

* update cpu

* add GPU

* add gpu

* update readme

README.md

@@ -205,6 +205,7 @@ Over 50 models have been optimized/verified on `ipex-llm`, including *LLaMA/LLaM
 | StableLM | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/stablelm) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/stablelm) |
 | CodeGemma | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/codegemma) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/codegemma) |
 | Command-R/cohere | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/cohere) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/cohere) |
+| CodeGeeX2 | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/codegeex2) | [link](python/llm/example/GPU/HF-Transformers-AutoModels/Model/codegeex2) |
 
 ## Get Support
 - Please report a bug or raise a feature request by opening a [Github Issue](https://github.com/intel-analytics/ipex-llm/issues)

docs/readthedocs/source/index.rst

@@ -604,6 +604,13 @@ Verified Models
          <td>
            <a href="https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/example/GPU/HF-Transformers-AutoModels/Model/cohere">link</a></td>
        </tr>
+       <tr>
+         <td>CodeGeeX2</td>
+         <td>
+           <a href="https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/example/CPU/HF-Transformers-AutoModels/Model/codegeex2">link</a></td>
+         <td>
+           <a href="https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/example/GPU/HF-Transformers-AutoModels/Model/codegeex2">link</a></td>
+       </tr>
      </tbody>
    </table>
 

python/llm/example/CPU/HF-Transformers-AutoModels/Model/codegeex2/README.md

@@ -0,0 +1,83 @@
+# CodeGeeX2
+
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on CodeGeex2 models which is implemented based on the ChatGLM2 architecture trained on more code data. We utilize the [THUDM/codegeex2-6b](https://huggingface.co/THUDM/codegeex2-6b) as a reference CodeGeeX2 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 1: Predict Tokens using `generate()` API
+In the example [generate.py](./generate.py), we show a basic use case for a CodeGeeX2 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 # 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
+```
+
+On Windows:
+
+```cmd
+conda create -n llm python=3.11
+conda activate llm
+
+pip install --pre --upgrade ipex-llm[all]
+```
+
+### 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 CodeGeex2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/codegeex2-6b'`.
+- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'# language: Python\n# write a bubble sort function\n'`.
+- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `128`.
+
+#### 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 -t
+
+# 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
+#### [THUDM/codegeex2-6b](https://huggingface.co/THUDM/codegeex2-6b)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+# language: Python
+# write a bubble sort function
+
+-------------------- Output --------------------
+# language: Python
+# write a bubble sort function
+
+
+def bubble_sort(lst):
+    for i in range(len(lst) - 1):
+        for j in range(len(lst) - 1 - i):
+            if lst[j] > lst[j + 1]:
+                lst[j], lst[j + 1] = lst[j + 1], lst[j]
+    return lst
+
+
+print(bubble_sort([1, 2, 3, 4, 5, 6, 7, 8,
+```

python/llm/example/CPU/HF-Transformers-AutoModels/Model/codegeex2/generate.py

@@ -0,0 +1,69 @@
+#
+# 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 ipex_llm.transformers import AutoModel
+from transformers import AutoTokenizer
+
+# you could tune the prompt based on your own model,
+# here the prompt tuning refers to https://huggingface.co/THUDM/codegeex2-6b#%E5%BF%AB%E9%80%9F%E5%BC%80%E5%A7%8B-%EF%BD%9C-get-started
+CODEGEEX_PROMPT_FORMAT = "{prompt}"
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for CodeGeeX2 model')
+    parser.add_argument('--repo-id-or-model-path', type=str, default="THUDM/codegeex2-6b",
+                        help='The huggingface repo id for the CodeGeeX2 model to be downloaded'
+                             ', or the path to the huggingface checkpoint folder')
+    parser.add_argument('--prompt', type=str, default="# language: Python\n# write a bubble sort function\n",
+                        help='Prompt to infer')
+    parser.add_argument('--n-predict', type=int, default=128,
+                        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 = AutoModel.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 = CODEGEEX_PROMPT_FORMAT.format(prompt=args.prompt)
+        input_ids = tokenizer.encode(prompt, return_tensors="pt")
+        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 IPEX-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)

python/llm/example/CPU/PyTorch-Models/Model/codegeex2/README.md

@@ -0,0 +1,83 @@
+# CodeGeeX2
+
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on CodeGeex2 models which is implemented based on the ChatGLM2 architecture trained on more code data. We utilize the [THUDM/codegeex2-6b](https://huggingface.co/THUDM/codegeex2-6b) as a reference CodeGeeX2 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 1: Predict Tokens using `generate()` API
+In the example [generate.py](./generate.py), we show a basic use case for a CodeGeeX2 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 # 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
+```
+
+On Windows:
+
+```cmd
+conda create -n llm python=3.11
+conda activate llm
+
+pip install --pre --upgrade ipex-llm[all]
+```
+
+### 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 CodeGeex2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/codegeex2-6b'`.
+- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'# language: Python\n# write a bubble sort function\n'`.
+- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `128`.
+
+#### 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 -t
+
+# 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
+#### [THUDM/codegeex2-6b](https://huggingface.co/THUDM/codegeex2-6b)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+# language: Python
+# write a bubble sort function
+
+-------------------- Output --------------------
+# language: Python
+# write a bubble sort function
+
+
+def bubble_sort(lst):
+    for i in range(len(lst) - 1):
+        for j in range(len(lst) - 1 - i):
+            if lst[j] > lst[j + 1]:
+                lst[j], lst[j + 1] = lst[j + 1], lst[j]
+    return lst
+
+
+print(bubble_sort([1, 2, 3, 4, 5, 6, 7, 8,
+```

python/llm/example/CPU/PyTorch-Models/Model/codegeex2/generate.py

@@ -0,0 +1,69 @@
+#
+# 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 ipex_llm.transformers import AutoModel
+from transformers import AutoTokenizer
+
+# you could tune the prompt based on your own model,
+# here the prompt tuning refers to https://huggingface.co/THUDM/codegeex2-6b#%E5%BF%AB%E9%80%9F%E5%BC%80%E5%A7%8B-%EF%BD%9C-get-started
+CODEGEEX_PROMPT_FORMAT = "{prompt}"
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for CodeGeeX2 model')
+    parser.add_argument('--repo-id-or-model-path', type=str, default="THUDM/codegeex2-6b",
+                        help='The huggingface repo id for the CodeGeeX2 model to be downloaded'
+                             ', or the path to the huggingface checkpoint folder')
+    parser.add_argument('--prompt', type=str, default="# language: Python\n# write a bubble sort function\n",
+                        help='Prompt to infer')
+    parser.add_argument('--n-predict', type=int, default=128,
+                        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 = AutoModel.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 = CODEGEEX_PROMPT_FORMAT.format(prompt=args.prompt)
+        input_ids = tokenizer.encode(prompt, return_tensors="pt")
+        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 IPEX-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)

python/llm/example/GPU/HF-Transformers-AutoModels/Model/codegeex2/README.md

@@ -0,0 +1,138 @@
+# CodeGeeX2
+
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on CodeGeeX2 models which is implemented based on the ChatGLM2 architecture trained on more code data on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [THUDM/codegeex-6b](https://huggingface.co/THUDM/codegeex2-6b) as a reference CodeGeeX2 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 1: Predict Tokens using `generate()` API
+In the example [generate.py](./generate.py), we show a basic use case for a CodeGeeX2 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/
+```
+
+#### 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/
+```
+
+### 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
+<details>
+
+<summary>For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series</summary>
+
+```bash
+export USE_XETLA=OFF
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export SYCL_CACHE_PERSISTENT=1
+```
+
+</details>
+
+<details>
+
+<summary>For Intel Data Center GPU Max Series</summary>
+
+```bash
+export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export 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`.
+</details>
+
+<details>
+
+<summary>For Intel iGPU</summary>
+
+```bash
+export SYCL_CACHE_PERSISTENT=1
+export BIGDL_LLM_XMX_DISABLED=1
+```
+
+</details>
+
+#### 3.2 Configurations for Windows
+<details>
+
+<summary>For Intel iGPU</summary>
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+set BIGDL_LLM_XMX_DISABLED=1
+```
+
+</details>
+
+<details>
+
+<summary>For Intel Arc™ A-Series Graphics</summary>
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+```
+
+</details>
+
+> [!NOTE]
+> For the first time that each model runs on Intel iGPU/Intel Arc™ A300-Series or Pro A60, it may take several minutes to compile.
+
+### 4. Running examples
+```
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
+```
+
+Arguments info:
+- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the CodeGeeX2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/codegeex-6b'`.
+- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'# language: Python\n# write a bubble sort function\n'`.
+- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `128`.
+
+#### Sample Output
+#### [THUDM/codegeex-6b](https://huggingface.co/THUDM/codegeex-6b)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+# language: Python
+# write a bubble sort function
+
+-------------------- Output --------------------
+# language: Python
+# write a bubble sort function
+
+
+def bubble_sort(lst):
+    for i in range(len(lst) - 1):
+        for j in range(len(lst) - 1 - i):
+            if lst[j] > lst[j + 1]:
+                lst[j], lst[j + 1] = lst[j + 1], lst[j]
+    return lst
+
+
+print(bubble_sort([5, 2, 3, 4, 1]))
+```

python/llm/example/GPU/HF-Transformers-AutoModels/Model/codegeex2/generate.py

@@ -0,0 +1,81 @@
+#
+# 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 ipex_llm.transformers import AutoModel
+from transformers import AutoTokenizer
+
+# you could tune the prompt based on your own model,
+# here the prompt tuning refers to https://huggingface.co/THUDM/codegeex2-6b#%E5%BF%AB%E9%80%9F%E5%BC%80%E5%A7%8B-%EF%BD%9C-get-started
+CODEGEEX_PROMPT_FORMAT = "{prompt}"
+
+
+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/codegeex2-6b",
+                        help='The huggingface repo id for the CodeGeeX2 model to be downloaded'
+                             ', or the path to the huggingface checkpoint folder')
+    parser.add_argument('--prompt', type=str, default="# language: Python\n# write a bubble sort function\n",
+                        help='Prompt to infer')
+    parser.add_argument('--n-predict', type=int, default=128,
+                        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
+    # When running LLMs on Intel iGPUs for Windows users, we recommend setting `cpu_embedding=True` in the from_pretrained function.
+    # This will allow the memory-intensive embedding layer to utilize the CPU instead of iGPU.
+    model = AutoModel.from_pretrained(model_path,
+                                      load_in_4bit=True,
+                                      optimize_model=True,
+                                      trust_remote_code=True,
+                                      use_cache=True)
+    model = model.half().to('xpu')
+
+    # Load tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(model_path,
+                                              trust_remote_code=True)
+
+    # Generate predicted tokens
+    with torch.inference_mode():
+        prompt = CODEGEEX_PROMPT_FORMAT.format(prompt=args.prompt)
+        input_ids = tokenizer.encode(prompt, return_tensors="pt").to('xpu')
+        # ipex_llm 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 IPEX-LLM INT4 optimizations
+        output = model.generate(input_ids,
+                                max_new_tokens=args.n_predict)
+        torch.xpu.synchronize()
+        end = time.time()
+        output_str = tokenizer.decode(output[0], skip_special_tokens=True)
+        print(f'Inference time: {end-st} s')
+        print('-'*20, 'Prompt', '-'*20)
+        print(prompt)
+        print('-'*20, 'Output', '-'*20)
+        print(output_str)

python/llm/example/GPU/PyTorch-Models/Model/codegeex2/README.md

@@ -0,0 +1,138 @@
+# CodeGeeX2
+
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on CodeGeeX2 models which is implemented based on the ChatGLM2 architecture trained on more code data on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [THUDM/codegeex-6b](https://huggingface.co/THUDM/codegeex2-6b) as a reference CodeGeeX2 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 1: Predict Tokens using `generate()` API
+In the example [generate.py](./generate.py), we show a basic use case for a CodeGeeX2 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/
+```
+
+#### 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/
+```
+
+### 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
+<details>
+
+<summary>For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series</summary>
+
+```bash
+export USE_XETLA=OFF
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export SYCL_CACHE_PERSISTENT=1
+```
+
+</details>
+
+<details>
+
+<summary>For Intel Data Center GPU Max Series</summary>
+
+```bash
+export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export 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`.
+</details>
+
+<details>
+
+<summary>For Intel iGPU</summary>
+
+```bash
+export SYCL_CACHE_PERSISTENT=1
+export BIGDL_LLM_XMX_DISABLED=1
+```
+
+</details>
+
+#### 3.2 Configurations for Windows
+<details>
+
+<summary>For Intel iGPU</summary>
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+set BIGDL_LLM_XMX_DISABLED=1
+```
+
+</details>
+
+<details>
+
+<summary>For Intel Arc™ A-Series Graphics</summary>
+
+```cmd
+set SYCL_CACHE_PERSISTENT=1
+```
+
+</details>
+
+> [!NOTE]
+> For the first time that each model runs on Intel iGPU/Intel Arc™ A300-Series or Pro A60, it may take several minutes to compile.
+
+### 4. Running examples
+```
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
+```
+
+Arguments info:
+- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the CodeGeeX2 model to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'THUDM/codegeex-6b'`.
+- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'# language: Python\n# write a bubble sort function\n'`.
+- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `128`.
+
+#### Sample Output
+#### [THUDM/codegeex-6b](https://huggingface.co/THUDM/codegeex-6b)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+# language: Python
+# write a bubble sort function
+
+-------------------- Output --------------------
+# language: Python
+# write a bubble sort function
+
+
+def bubble_sort(lst):
+    for i in range(len(lst) - 1):
+        for j in range(len(lst) - 1 - i):
+            if lst[j] > lst[j + 1]:
+                lst[j], lst[j + 1] = lst[j + 1], lst[j]
+    return lst
+
+
+print(bubble_sort([5, 2, 3, 4, 1]))
+```

python/llm/example/GPU/PyTorch-Models/Model/codegeex2/generate.py

@@ -0,0 +1,81 @@
+#
+# 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 ipex_llm.transformers import AutoModel
+from transformers import AutoTokenizer
+
+# you could tune the prompt based on your own model,
+# here the prompt tuning refers to https://huggingface.co/THUDM/codegeex2-6b#%E5%BF%AB%E9%80%9F%E5%BC%80%E5%A7%8B-%EF%BD%9C-get-started
+CODEGEEX_PROMPT_FORMAT = "{prompt}"
+
+
+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/codegeex2-6b",
+                        help='The huggingface repo id for the CodeGeeX2 model to be downloaded'
+                             ', or the path to the huggingface checkpoint folder')
+    parser.add_argument('--prompt', type=str, default="# language: Python\n# write a bubble sort function\n",
+                        help='Prompt to infer')
+    parser.add_argument('--n-predict', type=int, default=128,
+                        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
+    # When running LLMs on Intel iGPUs for Windows users, we recommend setting `cpu_embedding=True` in the from_pretrained function.
+    # This will allow the memory-intensive embedding layer to utilize the CPU instead of iGPU.
+    model = AutoModel.from_pretrained(model_path,
+                                      load_in_4bit=True,
+                                      optimize_model=True,
+                                      trust_remote_code=True,
+                                      use_cache=True)
+    model = model.half().to('xpu')
+
+    # Load tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(model_path,
+                                              trust_remote_code=True)
+
+    # Generate predicted tokens
+    with torch.inference_mode():
+        prompt = CODEGEEX_PROMPT_FORMAT.format(prompt=args.prompt)
+        input_ids = tokenizer.encode(prompt, return_tensors="pt").to('xpu')
+        # ipex_llm 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 IPEX-LLM INT4 optimizations
+        output = model.generate(input_ids,
+                                max_new_tokens=args.n_predict)
+        torch.xpu.synchronize()
+        end = time.time()
+        output_str = tokenizer.decode(output[0], skip_special_tokens=True)
+        print(f'Inference time: {end-st} s')
+        print('-'*20, 'Prompt', '-'*20)
+        print(prompt)
+        print('-'*20, 'Output', '-'*20)
+        print(output_str)