My Virtual Moments Repository

Now we have implemented the main functions of Virtual Moments.

The main entrance is at webpage/VM.html. To fully unlock the functions we have implemented, you shall run webpage/webapi.py on the remote server.

To run on your computer:

1. Clone the github repository by running on both your lab server and your local computer:

git clone git@github.com:EnderXie23/virtual-moments.git

2. On your lab server, run webpage/webapi.py:

cd ./virtual-moments/webpage
python webapi.py

3. On your local machine, double click to run webpage/VM.html.

To add your own virtual friend:

Let's say your virtual friend is named "Happy" (Note: We currently only support one word names only), follow the steps below to start chatting with him/her!

1. Add training data to the repository.

The data shall be in the following format:

# One line to represent a question
# One line to represent an answer
What would you say when it is raining?
Hmm... The rain drops are like jwelry falling from the sky!
...
...

Note that the comments shall not be present in your training data. No other information (Cues, hints, etc.) shall be present in your training data. About 25 Q&A examples is enough for training, but the data works best if they cover multiple realms of everyday life and can reflect the tone in which your friend speaks.

Now name your file as happy.txt and put it under the webpage/text/ folder. Your file name shall all be in lower case letters.

2. Add your friend's icon to the repository.

Add a image file under the directory webpage/photos/. There is no strict naming requirements for this file. Let's say this image file is named happy.png.

3. Do some changes to chatpage/js/chatscript.js.

On the first few lines of this script, you'll see:

const additionalCharacters = [
    // { character: "YOUR_CHARACTER", file: "../webpage/photos/YOUR_PNG.png", location: "YOUR_LOCATION" },
]

Now modify this line as:

const additionalCharacters = [
    { character: "Happy", file: "../webpage/photos/happy.png", location: "Beijing" },
]

The location variable is optional, which is used to beautify the interface.

4. All the changes shall take place on both your lab server and your local machine!

5. Re-run chatpage/ChatUI.html to chat with your new friend!

LOGS

Congratulations! Now the HTML page can correctly call all the functions we have implemented.

Update Virtual Moments HTML.

Update multiple character chatting interface.

Congratulations! Now we can use a webui to chat with the LLM with character! (Yet the contents now is limited to Furina)

TASKS

• Implement a LLM agent with memory so that it can read from file and generate answers in the same manner.
• Implement a script that can generate an HTML page.
• Implement an API service that can handle HTML requests.
• (Not going to do) Implement a script that may generate images according to context.