This is implementation of L2p for Continual learning. This uses two of the latest technique Viz, Prompt learning and Vision Transformer for continual learning.
To overcome the large constraints of Catastropic forgetting without using any mainstream solutions like Buffer Based, Architectural based, Regularization based which uses either large data buffer or test ID at test time which is highhly undeesirable. So we come up with a very efficient any low data requiring technique.
Which compelete Rehersal based technique(SOTA) without any buffer data.