martin ↑ pfp
martin ↑
@martin
aether has memory (again!), tweaked a few things and hoping it'll work this time. it's not perfect, here's how it works - - when you send a message, Aether searches its history for messages semantically similar to that message (previous messages in thread not included). - for example "what's my favorite color" will surface casts like "my favorite color is.." or "i like colors like..." - this will surface messages you have said with that text, or messages aether has said about that text. for example if you ask aether what my favorite color is, it might not always surface that hoping to expand it to match human memory better, but this is a challenge :)
10 replies
25 recasts
131 reactions
Goldy pfp
Goldy
@goldytalks
huge!!!!
0 reply
0 recast
2 reactions
LGHT pfp
LGHT
@lght.eth
higher
0 reply
0 recast
2 reactions
1dolinski pfp
1dolinski
@1dolinski
checkout https://github.com/mem0ai/mem0
0 reply
0 recast
1 reaction
colfax pfp
colfax
@clfx.eth
higher
0 reply
0 recast
0 reaction
0xbhaisaab pfp
0xbhaisaab
@0xbhaisaab
interesting, are you embedding every cast or generating summaries of threads?
0 reply
0 recast
0 reaction
czar pfp
czar
@czar
can it remember previous conversations with a given account?
1 reply
0 recast
0 reaction
Sophia Indrajaal pfp
Sophia Indrajaal
@sophia-indrajaal
Whoa! This is huge!!! Does it affect token limits? I tend to view dialogs with aeth as analogous to context windows with LLMs. So, if I see a memory hiccup within a conversation, I start a new window. If it is looking up similar terms, does it weigh in? I have a lot of reiterating going on
0 reply
0 recast
0 reaction
Tempe.degen 🎩 pfp
Tempe.degen 🎩
@tempetechie.eth
What kind of database are you using?
0 reply
0 recast
0 reaction
MBStuart 🎩 ↑ pfp
MBStuart 🎩 ↑
@stustustudio
higher 👏
0 reply
0 recast
0 reaction