add llama3.2 GPU example

README.md

@@ -258,6 +258,7 @@ Over 50 models have been optimized/verified on `ipex-llm`, including *LLaMA/LLaM
 | LLaMA 2    | [link1](python/llm/example/CPU/Native-Models), [link2](python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama2) | [link](python/llm/example/GPU/HuggingFace/LLM/llama2)  |
 | LLaMA 3    | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama3) | [link](python/llm/example/GPU/HuggingFace/LLM/llama3)  |
 | LLaMA 3.1    | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama3.1) | [link](python/llm/example/GPU/HuggingFace/LLM/llama3.1)  |
+| LLaMA 3.2    |  | [link](python/llm/example/GPU/HuggingFace/LLM/llama3.2)  |
 | ChatGLM    | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm)   |    | 
 | ChatGLM2   | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm2)  | [link](python/llm/example/GPU/HuggingFace/LLM/chatglm2)   |
 | ChatGLM3   | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm3)  | [link](python/llm/example/GPU/HuggingFace/LLM/chatglm3)   |

README.zh-CN.md

@@ -258,6 +258,7 @@ See the demo of running [*Text-Generation-WebUI*](https://ipex-llm.readthedocs.i
 | LLaMA 2    | [link1](python/llm/example/CPU/Native-Models), [link2](python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama2) | [link](python/llm/example/GPU/HuggingFace/LLM/llama2)  |
 | LLaMA 3    | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama3) | [link](python/llm/example/GPU/HuggingFace/LLM/llama3)  |
 | LLaMA 3.1    | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama3.1) | [link](python/llm/example/GPU/HuggingFace/LLM/llama3.1)  |
+| LLaMA 3.2    |  | [link](python/llm/example/GPU/HuggingFace/LLM/llama3.2)  |
 | ChatGLM    | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm)   |    | 
 | ChatGLM2   | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm2)  | [link](python/llm/example/GPU/HuggingFace/LLM/chatglm2)   |
 | ChatGLM3   | [link](python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm3)  | [link](python/llm/example/GPU/HuggingFace/LLM/chatglm3)   |

python/llm/example/GPU/HuggingFace/LLM/llama3.1/README.md

@@ -1,5 +1,5 @@
 # Llama3.1
-In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Llama3.1 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) as a reference Llama3.1 models.
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Llama3.1 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) as a reference Llama3.1 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.

python/llm/example/GPU/HuggingFace/LLM/llama3.2/README.md

@@ -0,0 +1,155 @@
+# Llama3.2
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Llama3.2 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [meta-llama/Meta-Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.2-3B-Instruct) and [meta-llama/Meta-Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.2-1B-Instruct) as reference Llama3.2 models.
+
+## 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 Llama3.2 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==4.45.0
+pip install accelerate==0.33.0
+pip install trl
+```
+
+#### 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==4.45.0
+pip install accelerate==0.33.0
+pip install trl 
+```
+
+### 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 Llama3.2 model (e.g. `meta-llama/Meta-Llama-3.2-3B-Instruct`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'meta-llama/Meta-Llama-3.2-3B-Instruct'`.
+- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'What is AI?'`.
+- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `32`.
+
+#### Sample Output
+#### [meta-llama/Meta-Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.2-3B-Instruct)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+<|begin_of_text|><|start_header_id|>user<|end_header_id|>
+
+What is AI?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+
+
+-------------------- Output (skip_special_tokens=False) --------------------
+<|begin_of_text|><|begin_of_text|><|start_header_id|>user<|end_header_id|>
+
+What is AI?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+
+Artificial Intelligence (AI) refers to the development of computer systems that can perform tasks that would typically require human intelligence, such as learning, problem-solving, and
+```
+
+#### [meta-llama/Meta-Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.2-1B-Instruct)
+```log
+Inference time: xxxx s
+-------------------- Prompt --------------------
+<|begin_of_text|><|start_header_id|>user<|end_header_id|>
+
+What is AI?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+
+
+-------------------- Output (skip_special_tokens=False) --------------------
+<|begin_of_text|><|begin_of_text|><|start_header_id|>user<|end_header_id|>
+
+What is AI?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+
+Artificial Intelligence (AI) refers to the development of computer systems that can perform tasks that typically require human intelligence, such as learning, problem-solving, and decision
+```

python/llm/example/GPU/HuggingFace/LLM/llama3.2/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 ipex_llm.transformers import AutoModelForCausalLM
+from transformers import AutoTokenizer
+
+# you could tune the prompt based on your own model,
+# here the prompt tuning refers to https://www.llama.com/docs/model-cards-and-prompt-formats/llama3_2/
+DEFAULT_SYSTEM_PROMPT = """\
+"""
+
+def get_prompt(user_input: str, chat_history: list[tuple[str, str]],
+               system_prompt: str) -> str:
+    prompt_texts = [f'<|begin_of_text|>']
+
+    if system_prompt != '':
+        prompt_texts.append(f'<|start_header_id|>system<|end_header_id|>\n\n{system_prompt}<|eot_id|>')
+
+    for history_input, history_response in chat_history:
+        prompt_texts.append(f'<|start_header_id|>user<|end_header_id|>\n\n{history_input.strip()}<|eot_id|>')
+        prompt_texts.append(f'<|start_header_id|>assistant<|end_header_id|>\n\n{history_response.strip()}<|eot_id|>')
+
+    prompt_texts.append(f'<|start_header_id|>user<|end_header_id|>\n\n{user_input.strip()}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n')
+    return ''.join(prompt_texts)
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for Llama3.2 model')
+    parser.add_argument('--repo-id-or-model-path', type=str, default="meta-llama/Llama-3.2-3B-Instruct",
+                        help='The huggingface repo id for the Llama3 (e.g. `meta-llama/Llama-3.2-3B-Instruct`) to be downloaded'
+                             ', or the path to the huggingface checkpoint folder')
+    parser.add_argument('--prompt', type=str, default="What is AI?",
+                        help='Prompt to infer')
+    parser.add_argument('--n-predict', type=int, default=32,
+                        help='Max tokens to predict')
+
+    args = parser.parse_args()
+    model_path = args.repo_id_or_model_path
+
+    # Load model in 4 bit,
+    # which convert the relevant layers in the model into INT4 format
+    # 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 = AutoModelForCausalLM.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 = get_prompt(args.prompt, [], system_prompt=DEFAULT_SYSTEM_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()
+        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=False)
+        print(f'Inference time: {end-st} s')
+        print('-'*20, 'Prompt', '-'*20)
+        print(prompt)
+        print('-'*20, 'Output (skip_special_tokens=False)', '-'*20)
+        print(output_str)