Hello folks,
One of the frustrating things working with LLMs in general is their context window - sometimes it feels like talking with someone with major memory issues.
This is worst when you want it to solve a problem. At some point it will start repeating itself in a loop.
I wonder if there’s a way to have it automatically “write down” what it has already tried, and any other important things that came up in conversation with the developer. I know memory is in beta, but it’s not automatic - you have to tell it to “write it down”, and even then it doesn’t always “remember” it when push comes to shove.
Would love to hear how you’ve been solving this for yourselves.
Thanks in advance!