Alan Langford on Nostr: npub1mp3r0…aqjxj I still think the actual best use case for LLM AI is content ...
npub1mp3r053ngat7495tmkeg50hpwd4j6skyl8959mtvd50x8v0669esfaqjxj (npub1mp3…qjxj) I still think the actual best use case for LLM AI is content filtering.
As for your specific case, it might be possible to write a Spam Assassin filter for those. Worst case you might need to write a plugin, but that seems pretty easy to characterize, unless they also dump 2K of Bayes-busting text.
It's the Bayes busting where LLM AIs could really help out.
Published at
2023-08-09 13:05:20Event JSON
{
"id": "8e2992774d5b4e808a1c09ee15963a3e8ce27c7f035127f8f9e1654da40a0e79",
"pubkey": "cc80529453ad7aa4de01e8494d0232b5c2ad597dbf908f55dc291fa288601e04",
"created_at": 1691586320,
"kind": 1,
"tags": [
[
"p",
"d86237d2334757ea968bddb28a3ee1736b2d42c4f9cb42ed6c6d1e63b1fad173",
"wss://relay.mostr.pub"
],
[
"p",
"fc2e276b576a6b78e0b67094fc807c17c6065707fb0c1251c0c792b876c953ff",
"wss://relay.mostr.pub"
],
[
"e",
"e3ef51cd592cca525f536178747f07cb071310d8d6e9e80722987954b257cde1",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://mindly.social/users/alan/statuses/110859801123104756",
"activitypub"
]
],
"content": "nostr:npub1mp3r053ngat7495tmkeg50hpwd4j6skyl8959mtvd50x8v0669esfaqjxj I still think the actual best use case for LLM AI is content filtering.\n\nAs for your specific case, it might be possible to write a Spam Assassin filter for those. Worst case you might need to write a plugin, but that seems pretty easy to characterize, unless they also dump 2K of Bayes-busting text.\n\nIt's the Bayes busting where LLM AIs could really help out.",
"sig": "45cd54086f14f7dae99e4de9e2cd6f4bcd7fd5b9760b9aac54d5fba44f7cb5bc913123533db7e2821b8d2221638e1086dc76244fabc1c1aff5735435b4fbab3d"
}