search

VedalAI / neuro-amongus

Among Us Plugin for Neuro-sama
GNU General Public License v3.0
531 stars 50 forks source link

(Question) about dataset to help train the AI #79

Closed andresnowak closed 1 year ago

andresnowak commented 1 year ago

I have a question regarding contributing to the training of our AI model, specifically in relation to the neural network and dataset:

  1. How can someone contribute to the training of the AI model, particularly in the area of the neural network?
  2. Where can I access the dataset required for training the AI model? Or If someone wants to help in this area they should use the recordings from their own games for the dataset?
Alexejhero commented 1 year ago
  1. The model is trained by vedal using the data collected and provided by various people using the mod. In that regard, there is no way to help with the actual training, but if you have any contributions that you would like to make to the neural network architecture, the source code for it is available in this repository in the AI folder.
  2. Currently the dataset is not publicly accessible, as it is only stored locally on the upload server. As a result of that, only vedal has access to it and can use it to train the model. That being said, we believe that we have already collected all of the data that we need, so we have paused the collection process. As such, none of the new recordings uploaded to the server will be used for training.
andresnowak commented 1 year ago

Okay i understand, and thank you for answering.