Replies: 2 comments
-
|
@nicoloboschi @benfrank241 just in case you missed this, I do think these suggested changes would allow hindsight to scale beyond the current "memory for agent" current product positioning. Happy to discuss. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Hey thanks for the writeup! Feel free to open up separate PRs and we will have a look. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello people,
I've been going through the hindsight codebase quite extensively to figure out how well it suits what I'm trying to achieve.
I've formed a very opinion that some design decisions may have been taken with "agent memory" goal in mind, for instance "decay window" and "observation trend window".
I've prepared the attached plan that I would be interested in contributing if you'll have it.
TLDR: essentially extending some of the configs
_CONFIGURABLE_FIELDS, adding storage and retrieval capabilities, enriching some aspects of the data stored.Are you able to review the attached document and confirm whether you would
I don't think this is "major rework", but I would also want someone on your end to properly vet the resulting work.. You definitely know your thing better than I do, and likewise Postgres..
Not sure on my timeframe at this stage. If interested, I'm happy to work with someone on your side to get this out.
PRD-Phase2-Hindsight-Patches copy.md
Beta Was this translation helpful? Give feedback.
All reactions