It feels like having the memories explicitly stored by the agent itself as natural language is quite limiting. Would it be possible to add one or several memor

It feels like having the memories explicitly stored by the agent itself as natural language is quite limiting. Would it be possible to add one or several memory components to the LLM model itself while training it ? Example: https://en.wikipedia.org/wiki/Differentiable_neural_computer https://en.wikipedia.org/wiki/Differentiable_neural_computer I guess doing so would make the training less parallelizable ?