Made Custom AI Assistant fit the kit template (#162)

* Updated dependencies

* Added main script

* Using llama 3.2 in custom AI assistant

* Changed public argument

* Fixed chat model returned path

* Added log

Author: adrianboguszewski

ai_ref_kits/custom_ai_assistant/README.md

@@ -39,7 +39,7 @@ Now, let's dive into the steps starting with installing Python. We recommend usi
 
 ## Installing Prerequisites
 
-This project requires Python 3.8 or higher and a few libraries. If you don't have Python installed on your machine, go to https://www.python.org/downloads/ and download the latest version for your operating system. Follow the prompts to install Python, making sure to check the option to add Python to your PATH environment variable.
+This project requires Python 3.10 or higher and a few libraries. If you don't have Python installed on your machine, go to https://www.python.org/downloads/ and download the latest version for your operating system. Follow the prompts to install Python, making sure to check the option to add Python to your PATH environment variable.
 
 Install libraries and tools:
 
@@ -178,7 +178,7 @@ Execute the `app.py` script with the following command, including all necessary
 ```shell
 python app.py --asr_model_dir path/to/asr_model --chat_model_dir path/to/chat_model
 ```
-Replace `path/to/asr_model` and `path/to/chat_model` with actual paths to your respective models. Add `--public_interface` to make it publicly accessible.
+Replace `path/to/asr_model` and `path/to/chat_model` with actual paths to your respective models. Add `--public` to make it publicly accessible.
 
 ### Accessing the Web Interface
 After running the script, Gradio will provide a local URL, typically `http://127.0.0.1:XXXX`, which you can open in your web browser to start interacting with the assistant. If you configured the application to be accessible publicly, Gradio will also provide a public URL.

ai_ref_kits/custom_ai_assistant/app.py

@@ -323,15 +323,17 @@ def run(asr_model_dir: Path, chat_model_dir: Path, public_interface: bool = Fals
 
     # create user interface
     demo = create_UI(initial_message)
+
+    log.info("Demo is ready!")
     # launch demo
     demo.queue().launch(share=public_interface)
 
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
     parser.add_argument('--asr_model_dir', type=str, default="model/distil-whisper-large-v3-FP16", help="Path to the automatic speech recognition model directory")
-    parser.add_argument('--chat_model_dir', type=str, default="model/llama3.1-8B-INT4", help="Path to the chat model directory")
-    parser.add_argument('--public_interface', default=False, action="store_true", help="Whether interface should be available publicly")
+    parser.add_argument('--chat_model_dir', type=str, default="model/llama3.2-3B-INT4", help="Path to the chat model directory")
+    parser.add_argument('--public', default=False, action="store_true", help="Whether interface should be available publicly")
 
     args = parser.parse_args()
     run(Path(args.asr_model_dir), Path(args.chat_model_dir), args.public_interface)

ai_ref_kits/custom_ai_assistant/convert_and_optimize_chat.py

@@ -21,7 +21,7 @@ def convert_chat_model(model_type: str, precision: str, model_dir: Path) -> Path
         precision: model precision
         model_dir: dir to export model
     Returns:
-       Path to exported model
+       Path to exported model dir
     """
     output_dir = model_dir / model_type
     model_name = MODEL_MAPPING[model_type]
@@ -52,13 +52,13 @@ def convert_chat_model(model_type: str, precision: str, model_dir: Path) -> Path
     tokenizer = AutoTokenizer.from_pretrained(model_name)
     tokenizer.save_pretrained(output_dir)
 
-    return Path(output_dir) / "openvino_model.xml"
+    return Path(output_dir)
 
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
     parser.add_argument("--chat_model_type", type=str, choices=["llama3.1-8B", "llama3-8B", "qwen2-7B", "llama3.2-3B"],
-                        default="llama3.1-8B", help="Chat model to be converted")
+                        default="llama3.2-3B", help="Chat model to be converted")
     parser.add_argument("--precision", type=str, default="int4", choices=["fp16", "int8", "int4"], help="Model precision")
     parser.add_argument("--model_dir", type=str, default="model", help="Directory to place the model in")
 

ai_ref_kits/custom_ai_assistant/main.py

@@ -0,0 +1,28 @@
+import argparse
+from pathlib import Path
+
+import app
+import convert_and_optimize_asr as asr
+import convert_and_optimize_chat as chat
+
+
+def main(args):
+    asr_model_dir = asr.convert_asr_model(args.asr_model_type, args.asr_precision, Path(args.model_dir))
+    chat_model_dir = chat.convert_chat_model(args.chat_model_type, args.chat_precision, Path(args.model_dir))
+
+    app.run(asr_model_dir, chat_model_dir, args.public)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument("--asr_model_type", type=str, choices=["distil-whisper-large-v3", "belle-distilwhisper-large-v2-zh"],
+                        default="distil-whisper-large-v3", help="Speech recognition model to be converted")
+    parser.add_argument("--asr_precision", type=str, default="fp16", choices=["fp16", "int8"], help="ASR model precision")
+    parser.add_argument("--chat_model_type", type=str, choices=["llama3.1-8B", "llama3-8B", "qwen2-7B", "llama3.2-3B"],
+                        default="llama3.2-3B", help="Chat model to be converted")
+    parser.add_argument("--chat_precision", type=str, default="int4", choices=["fp16", "int8", "int4"], help="Chat model precision")
+    parser.add_argument("--model_dir", type=str, default="model", help="Directory to place the model in")
+    parser.add_argument('--public', default=False, action="store_true", help="Whether interface should be available publicly")
+
+    main(parser.parse_args())

ai_ref_kits/custom_ai_assistant/requirements.txt

@@ -1,16 +1,17 @@
 --extra-index-url https://download.pytorch.org/whl/cpu
 
-openvino==2024.4.0
-optimum-intel==1.19.0
-optimum==1.22.0
-nncf==2.13.0
+openvino==2024.6.0
+optimum-intel==1.21.0
+optimum==1.23.3
+nncf==2.14.1
 
 # onnx>1.16.1 doesn't work on windows
-onnx==1.16.1
+onnx==1.16.1; platform_system == "Windows"
+onnx==1.17.0; platform_system != "Windows"
 onnxruntime==1.17.3
-torch==2.4.1
+torch==2.5.1
 
-transformers==4.44.2
+transformers==4.46.3
 librosa==0.10.2
 
-gradio==5.7.1
+gradio==5.10.0