Add --modelscope in GPU examples for minicpm, minicpm3, baichuan2 (#12564)

* Add --modelscope for more models

* minicpm

---------

Co-authored-by: ATMxsp01 <shou.xu@intel.com>

python/llm/example/GPU/HuggingFace/LLM/baichuan2/README.md

@@ -1,5 +1,5 @@
 # Baichuan
-In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Baichuan2 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [baichuan-inc/Baichuan2-7B-Chat](https://huggingface.co/baichuan-inc/Baichuan-7B-Chat) as a reference Baichuan model.
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Baichuan2 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [baichuan-inc/Baichuan2-7B-Chat](https://huggingface.co/baichuan-inc/Baichuan-7B-Chat) (or [baichuan-inc/Baichuan2-7B-Chat](https://www.modelscope.cn/models/[baichuan-inc/Baichuan2-7B-Chat]) for ModelScope) as a reference Baichuan 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.
@@ -16,6 +16,9 @@ conda activate llm
 pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
 
 pip install transformers_stream_generator  # additional package required for Baichuan-7B-Chat to conduct generation
+
+# [optional] only needed if you would like to use ModelScope as model hub
+pip install modelscope==1.11.0
 ```
 
 #### 1.2 Installation on Windows
@@ -28,6 +31,9 @@ conda activate llm
 pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
 
 pip install transformers_stream_generator  # additional package required for Baichuan-7B-Chat to conduct generation
+
+# [optional] only needed if you would like to use ModelScope as model hub
+pip install modelscope==1.11.0
 ```
 
 ### 2. Configures OneAPI environment variables for Linux
@@ -95,14 +101,19 @@ set SYCL_CACHE_PERSISTENT=1
 > 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
 
-```
+```bash
+# for Hugging Face model hub
 python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
+
+# for ModelScope model hub
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT --modelscope
 ```
 
 Arguments info:
-- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the Baichuan model (e.g `baichuan-inc/Baichuan2-7B-Chat`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'baichuan-inc/Baichuan2-7B-Chat'`.
+- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the **Hugging Face** or **ModelScope** repo id for the Baichuan model (e.g `baichuan-inc/Baichuan2-7B-Chat`) to be downloaded, or the path to the checkpoint folder. It is default to be `'baichuan-inc/Baichuan2-7B-Chat'`.
 - `--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`.
+- `--modelscope`: using **ModelScope** as model hub instead of **Hugging Face**.
 
 #### Sample Output
 #### [baichuan-inc/Baichuan2-7B-Chat](https://huggingface.co/baichuan-inc/Baichuan2-7B-Chat)

python/llm/example/GPU/HuggingFace/LLM/baichuan2/generate.py

@@ -19,7 +19,6 @@ import time
 import argparse
 
 from ipex_llm.transformers import AutoModelForCausalLM
-from transformers import AutoTokenizer
 
 # prompt format referred from https://github.com/baichuan-inc/Baichuan2/issues/227 
 # and https://huggingface.co/baichuan-inc/Baichuan2-7B-Chat/blob/main/generation_utils.py#L7-L49
@@ -29,14 +28,24 @@ BAICHUAN_PROMPT_FORMAT = "<reserved_106> {prompt} <reserved_107>"
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for Baichuan model')
     parser.add_argument('--repo-id-or-model-path', type=str, default="baichuan-inc/Baichuan2-7B-Chat",
-                        help='The huggingface repo id for the Baichuan model to be downloaded'
+                        help='The Hugging Face repo id for the Baichuan model to be downloaded'
-                             ', or the path to the huggingface checkpoint folder')
+                             ', or the path to the 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')
+    parser.add_argument('--modelscope', action="store_true", default=False, 
+                        help="Use models from modelscope")
 
     args = parser.parse_args()
+
+    if args.modelscope:
+        from modelscope import AutoTokenizer
+        model_hub = 'modelscope'
+    else:
+        from transformers import AutoTokenizer
+        model_hub = 'huggingface'
+
     model_path = args.repo_id_or_model_path
 
     # Load model in 4 bit,
@@ -50,7 +59,8 @@ if __name__ == '__main__':
     model = AutoModelForCausalLM.from_pretrained(model_path,
                                                  load_in_4bit=True,
                                                  trust_remote_code=True,
-                                                 use_cache=True)
+                                                 use_cache=True,
+                                                 model_hub=model_hub)
     model = model.half().to('xpu')
 
     # Load tokenizer

python/llm/example/GPU/HuggingFace/LLM/minicpm/README.md

@@ -1,5 +1,5 @@
 # MiniCPM
-In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on MiniCPM models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [openbmb/MiniCPM-2B-sft-bf16](https://huggingface.co/openbmb/MiniCPM-2B-sft-bf16) as a reference MiniCPM model.
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on MiniCPM models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [openbmb/MiniCPM-2B-sft-bf16](https://huggingface.co/openbmb/MiniCPM-2B-sft-bf16) and [openbmb/MiniCPM-1B-sft-bf16](https://huggingface.co/openbmb/MiniCPM-1B-sft-bf16) (or [OpenBMB/MiniCPM-2B-sft-bf16](https://www.modelscope.cn/models/OpenBMB/MiniCPM-2B-sft-bf16) and [OpenBMB/MiniCPM-1B-sft-bf16](https://www.modelscope.cn/models/OpenBMB/MiniCPM-1B-sft-bf16) for ModelScope) as a reference MiniCPM 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.
@@ -15,6 +15,9 @@ 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>=4.36"
+
+# [optional] only needed if you would like to use ModelScope as model hub
+pip install modelscope==1.11.0
 ```
 
 #### 1.2 Installation on Windows
@@ -26,6 +29,9 @@ 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>=4.36"
+
+# [optional] only needed if you would like to use ModelScope as model hub
+pip install modelscope==1.11.0
 ```
 
 ### 2. Configures OneAPI environment variables for Linux
@@ -93,14 +99,19 @@ set SYCL_CACHE_PERSISTENT=1
 > 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
 
-```
+```bash
-python ./generate.py --prompt 'What is AI?'
+# for Hugging Face model hub
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
+
+# for ModelScope model hub
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT --modelscope
 ```
 
 Arguments info:
-- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the MiniCPM model (e.g. `openbmb/MiniCPM-2B-sft-bf16`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'openbmb/MiniCPM-2B-sft-bf16'`.
+- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the **Hugging Face** or **ModelScope** repo id for the MiniCPM model (e.g. `openbmb/MiniCPM-2B-sft-bf16` or `openbmb/MiniCPM-1B-sft-bf16`) to be downloaded, or the path to the checkpoint folder. It is default to be `'openbmb/MiniCPM-2B-sft-bf16'` for **Hugging Face** and `'OpenBMB/MiniCPM-2B-sft-bf16'` for **ModelScope**.
 - `--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`.
+- `--modelscope`: using **ModelScope** as model hub instead of **Hugging Face**.
 
 #### Sample Output
 #### [openbmb/MiniCPM-2B-sft-bf16](https://huggingface.co/openbmb/MiniCPM-2B-sft-bf16)
@@ -112,3 +123,12 @@ Inference time: xxxx s
 -------------------- Output --------------------
 <s> <用户>what is AI?<AI> AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It is a field of computer science
 ```
+
+#### [openbmb/MiniCPM-1B-sft-bf16](https://huggingface.co/openbmb/MiniCPM-1B-sft-bf16)
+
+```log
+-------------------- Prompt --------------------
+<用户>What is AI?<AI>
+-------------------- Output --------------------
+<s> <用户>What is AI?<AI> Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It involves the development of computer systems that
+```

python/llm/example/GPU/HuggingFace/LLM/minicpm/generate.py

@@ -19,22 +19,32 @@ import time
 import argparse
 
 from ipex_llm.transformers import AutoModelForCausalLM
-from transformers import AutoTokenizer
 
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for MiniCPM model')
-    parser.add_argument('--repo-id-or-model-path', type=str, default="openbmb/MiniCPM-2B-sft-bf16",
+    parser.add_argument('--repo-id-or-model-path', type=str,
-                        help='The huggingface repo id for the MiniCPM model to be downloaded'
+                        help='The Hugging Face or ModelScope repo id for the MiniCPM model to be downloaded'
-                             ', or the path to the huggingface checkpoint folder')
+                             ', or the path to the 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')
+    parser.add_argument('--modelscope', action="store_true", default=False, 
+                        help="Use models from modelscope")
 
     args = parser.parse_args()
-    model_path = args.repo_id_or_model_path
 
+    if args.modelscope:
+        from modelscope import AutoTokenizer
+        model_hub = 'modelscope'
+    else:
+        from transformers import AutoTokenizer
+        model_hub = 'huggingface'
+    
+    model_path = args.repo_id_or_model_path if args.repo_id_or_model_path else \
+        ("OpenBMB/MiniCPM-2B-sft-bf16" if args.modelscope else "openbmb/MiniCPM-2B-sft-bf16")
+    
     # 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.
@@ -43,9 +53,10 @@ if __name__ == '__main__':
                                                  load_in_4bit=True,
                                                  trust_remote_code=True,
                                                  optimize_model=True,
-                                                 use_cache=True)
+                                                 use_cache=True,
+                                                 model_hub=model_hub)
     
-    model = model.to('xpu')
+    model = model.half().to('xpu')
 
     # Load tokenizer
     tokenizer = AutoTokenizer.from_pretrained(model_path,

python/llm/example/GPU/HuggingFace/LLM/minicpm3/README.md

@@ -1,5 +1,5 @@
 # MiniCPM3
-In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on MiniCPM3 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [openbmb/MiniCPM3-4B](https://huggingface.co/openbmb/MiniCPM3-4B) as a reference MiniCPM3 model.
+In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on MiniCPM3 models on [Intel GPUs](../../../README.md). For illustration purposes, we utilize the [openbmb/MiniCPM3-4B](https://huggingface.co/openbmb/MiniCPM3-4B) (or [OpenBMB/MiniCPM3-4B](https://www.modelscope.cn/models/OpenBMB/MiniCPM3-4B) for ModelScope) as a reference MiniCPM3 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.
@@ -16,6 +16,9 @@ conda activate llm
 pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
 
 pip install jsonschema datamodel_code_generator
+
+# [optional] only needed if you would like to use ModelScope as model hub
+pip install modelscope==1.11.0
 ```
 
 #### 1.2 Installation on Windows
@@ -28,6 +31,9 @@ conda activate llm
 pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
 
 pip install jsonschema datamodel_code_generator
+
+# [optional] only needed if you would like to use ModelScope as model hub
+pip install modelscope==1.11.0
 ```
 
 ### 2. Configures OneAPI environment variables for Linux
@@ -95,14 +101,19 @@ set SYCL_CACHE_PERSISTENT=1
 > 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
 
-```
+```bash
-python ./generate.py --prompt 'What is AI?'
+# for Hugging Face model hub
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
+
+# for ModelScope model hub
+python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT --modelscope
 ```
 
 Arguments info:
-- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the MiniCPM3 model (e.g. `openbmb/MiniCPM3-4B`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'openbmb/MiniCPM3-4B'`.
+- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the **Hugging Face** or **ModelScope** repo id for the MiniCPM3 model (e.g. `openbmb/MiniCPM3-4B`) to be downloaded, or the path to the checkpoint folder. It is default to be `'openbmb/MiniCPM3-4B'` for **Hugging Face** or `'OpenBMB/MiniCPM3-4B'` for **ModelScope**.
 - `--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`.
+- `--modelscope`: using **ModelScope** as model hub instead of **Hugging Face**.
 
 #### Sample Output
 #### [openbmb/MiniCPM3-4B](https://huggingface.co/openbmb/MiniCPM3-4B)

python/llm/example/GPU/HuggingFace/LLM/minicpm3/generate.py

@@ -19,21 +19,31 @@ import time
 import argparse
 
 from ipex_llm.transformers import AutoModelForCausalLM
-from transformers import AutoTokenizer
 
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='Predict Tokens using `generate()` API for MiniCPM3 model')
-    parser.add_argument('--repo-id-or-model-path', type=str, default="openbmb/MiniCPM3-4B",
+    parser.add_argument('--repo-id-or-model-path', type=str, 
-                        help='The huggingface repo id for the MiniCPM3 model to be downloaded'
+                        help='The Hugging Face or ModelScope repo id for the MiniCPM3 model to be downloaded'
-                             ', or the path to the huggingface checkpoint folder')
+                             ', or the path to the 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')
+    parser.add_argument('--modelscope', action="store_true", default=False, 
+                        help="Use models from modelscope")
 
     args = parser.parse_args()
-    model_path = args.repo_id_or_model_path
+    
+    if args.modelscope:
+        from modelscope import AutoTokenizer
+        model_hub = 'modelscope'
+    else:
+        from transformers import AutoTokenizer
+        model_hub = 'huggingface'
+    
+    model_path = args.repo_id_or_model_path if args.repo_id_or_model_path else \
+        ("OpenBMB/MiniCPM3-4B" if args.modelscope else "openbmb/MiniCPM3-4B")
 
     # Load model in 4 bit,
     # which convert the relevant layers in the model into INT4 format
@@ -43,7 +53,8 @@ if __name__ == '__main__':
                                                  load_in_4bit=True,
                                                  trust_remote_code=True,
                                                  optimize_model=True,
-                                                 use_cache=True)
+                                                 use_cache=True,
+                                                 model_hub=model_hub)
     
     model = model.half().to('xpu')