User:JonathanDocuments/sandbox

DataPredict (or formerly known as Aqwam's Roblox Machine And Deep Learning Library) is a machine and deep learning for Roblox. It contains more than twelve models and a variety of features. Its API is also beginner-friendly at the cost of customizability. The library also includes a set of API design standards if users wishes to create their own models and optimizers.

AR-MachineLL
The prototype version of the model was first made on January 2023. It was named as Aqwam's Roblox Machine Learning Library (or as known as AR-MachineLL). It was published to the public in Roblox's DevForum and it was well received prior the post's deletion. It had over 50 likes and 2.1K views. The original post was deleted due to the original creator stated it was an outdated library and had API design and calculation issues.

DataPredict
DataPredict was released on the same month to replace the prototype version of the library. DataPredict had a significant improvement over the prototype due to its object-oriented design. Initially, it was not well received due to lack of documentation, to which the creator had apologized to the community and made improvements over time.

Friendly API
The users do not have to build the models from scratch. Instead, it comes with fully working models and optimizers containing all the functions necessary for training.

Retrainable Models
The models can be retrained as many times the user wishes. This makes the library an excellent choice if the users want to have adaptable models.

Model Parameters Saving And Loading
The users can get the model parameters and save them into data storeages such as Roblox's DataStore service. The model parameters can be loaded back to the model once the model settings have been set to the original one that it had been trained on.

Sequential Model Poor Performance
The LSTM and RNN models takes too long to train. This is mainly due to sequential nature of the models.

Calculation Errors
Some models may calculate wrong model parameters values. This is due to the fact that the creator's weak mathematical skills. Nevertheless, these often get improved over time.