English | 简体中文
DSN is a powerful and innovative interactive prompt framework that combines the power of large language models with deep learning capabilities, creating truly intelligent and engaging conversational experiences.
DSN can:
- Understand and respond to your natural language requests - Ask it anything, and it will do its best to fulfill your requests.
- Execute commands and scripts - The framework will automatically execute code or commands on your computer, using actions to directly handle your requests.
- Search files and information - Effortlessly find files on your computer or search for information online.
- Learn from your interactions - DSN continuously learns from your interactions, improving its responses over time.
- Integrate with external plugins - Easily extend DSN's functionality using custom plugins.
- Support local model deployment - Run DSN offline using your custom models.
- Generate experience-accelerated models - Leverage your previous conversation history to build faster and more efficient models.
- More features are waiting to be explored!
Features:
- Voice control: Speak to DSN and let it hear your commands!
- Local model support: Use your own custom models for privacy and offline access.
- Advanced search capabilities: Search files by name, type, or keyword.
- Customizable settings: Adjust DSN to your needs by modifying settings in
config.py
. - Error handling: DSN handles errors gracefully and offers fully autonomous troubleshooting.
- Experience acceleration: Train DSN with your chat history for increased efficiency and speed.
Getting started:
1. Prerequisites:
- Python 3.11+: Install Python from the official website (version 3.11.2 is recommended).
- Google Cloud Platform API Key (for online model access): You can get a free trial API key here.
- Everything Search Engine (for file searching): Download the full version from its website and install it in the "binaries" folder of this cloned repository.
- Moondream model: Download from Huggingface and place it in the
instances\\moondream
folder of this cloned repository. - ChatTTS model: Obtain the modified integrated ChatTTS model from here, then unzip it into the
instances\\chatTTS
folder of this cloned repository. Extraction code: 1145 - Paraformer model: Obtain it from Modelscope, then unzip it into the
instances\\paraformer-zh
folder of this cloned repository. - Note: When unzipping the above models, simply unzip the contained files into the specified folder, do not unzip the folder within the compressed file.
2. Installation:
- Clone the repository:
git clone https://github.com/ccjjfdyqlhy/DSN.git
- Install necessary packages from
install_before_requirements.txt
. - Download the DSN-local.NT model:
- This model is not currently available publicly. Please contact the author to obtain it personally.
- Install dependencies:
pip install -r requirements.txt
3. Configuration:
-
Open
config.py
and adjust the following settings:- GENAI_APIKEY: Your Google Cloud Platform API key.
- USERNAME: Your desired username.
- CITY: Your city.
- TIMEZONE: Your preferred timezone.
- USER_PYTHON: The path to your preferred Python interpreter.
- AI_NAME: The name you want to give your AI.
- Set
MODEL_PATH
to the absolute path of theDSN-local.NT
file.
-
Optional:
- You can enable voice control by setting
SPEECH_CONTROL
toTrue
inconfig.py
. - For offline use, set
USE_LOCAL
toTrue
.
- You can enable voice control by setting
-
For more information, please refer to
config.py
.
4. Launching DSN:
- Open a command prompt or terminal and navigate to the project directory:
cd DSN
- Run the main program:
python DSN.Launch.py
5. Enjoy the conversation!
DSN is now ready to answer your questions, execute your commands, and help you with your tasks.
How to build an experience-accelerated model:
- Start a conversation with DSN.
- Type "BUILD_TRAIN_DATA".
- DSN will guide you to read the conversation history files in the
generated\\chat_history
folder. - After reading each file, summarize the key points and provide your feedback.
- Once satisfied, DSN will build an experience-accelerated model based on your input.
Other Notes:
- Please Note: You must read and agree to all terms of use in the
LICENCE
file before starting to use. - You will be presented with other features upon activating the interactive terminal.