Sometimes you can clearly see that the enhanced prompt is not quite what you want. Currently the prompt is fed straight into the generation without stopping. This can lead to either bad generations or fully cancelling the generation and trying again from scratch.
There should be an option (toggleable) to have a step in between where you can make manual changes to the enhancement. Being fully dependent on the LLM with no way to correct small mistakes or misinterpretations makes enhancement an all-or-nothing proposition. This can be especially pertinent when you only need to change a couple words here or there, rather than the entire text being bad.
Options to regenerate would also help with learning how the enhancer interprets things to help with having better initial prompts to begin with.
Sometimes you can clearly see that the enhanced prompt is not quite what you want. Currently the prompt is fed straight into the generation without stopping. This can lead to either bad generations or fully cancelling the generation and trying again from scratch.
There should be an option (toggleable) to have a step in between where you can make manual changes to the enhancement. Being fully dependent on the LLM with no way to correct small mistakes or misinterpretations makes enhancement an all-or-nothing proposition. This can be especially pertinent when you only need to change a couple words here or there, rather than the entire text being bad.
Options to regenerate would also help with learning how the enhancer interprets things to help with having better initial prompts to begin with.