Instruct mode loses the context of the conversation with every instruction

[tjohnman]

Alpaca models are essentially one-shot only. The user cannot refer to any of the previous information they or the model have provided. Each instruction is essentially independent from the rest. I don't know if this is a limitation of how the model is trained or whether this has been discussed here before (or if this is intentional).

It would be useful to have an experience more similar to ChatGPT, in which the context can be built up with a back and forth with the model and make the interaction more useful.

[tjohnman]

I'm sorry, I think this is no longer the case. I don't see a way to delete this issue though.