stovis on Nostr: Why is it that LLMs, which after all do some kind of statistical analyisis and number ...
Why is it that LLMs, which after all do some kind of statistical analyisis and number crunching, are not deterministic/consistent? Meaning they can't reliably produce the same output from the same prompt, or make small adjustments to a part of the output that doesn't affect the rest of the output?
#AskFedi #AI
Published at
2026-07-31 05:41:37 UTCEvent JSON
{
"id": "392f135f15c1727fcd89a32e79f0f5d22f7f156e637bb677e8f81c7856e07eda",
"pubkey": "f1ff90c1da598c5e2a19e528350ad29bc048794af880b9c8f0fc5bb1e5ba6da4",
"created_at": 1785476497,
"kind": 1,
"tags": [
[
"t",
"ai"
],
[
"proxy",
"https://fosstodon.org/@stovis/117012987745472410",
"web"
],
[
"t",
"askfedi"
],
[
"proxy",
"https://fosstodon.org/users/stovis/statuses/117012987745472410",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://fosstodon.org/users/stovis/statuses/117012987745472410",
"pink.momostr"
],
[
"-"
]
],
"content": "Why is it that LLMs, which after all do some kind of statistical analyisis and number crunching, are not deterministic/consistent? Meaning they can't reliably produce the same output from the same prompt, or make small adjustments to a part of the output that doesn't affect the rest of the output?\n\n#AskFedi #AI",
"sig": "670954fe29740d3850600a5611d93153c60a89f12925f722c5e588eb88a5f8d1d24d3984fd4b9498cae15ca6e5c053a78c80ddee24984206728541e5f453365d"
}
