tante on Nostr: Question: Do you know of a paper analyzing whether using a markov chain (as used in ...
Question: Do you know of a paper analyzing whether using a markov chain (as used in many AI crawler tarpits) to generate training data does lead to the same/a smaller/a bigger level of model collapse than generating training data with an LLM? #ai
Published at
2025-01-22 20:46:49Event JSON
{
"id": "70509614e327a8cac3fc27970ff5329eddd9935780d53da6610e261cdfedf636",
"pubkey": "6e5bb3f4df765d024066fb8b1b0ab196b68726b8092782a387c9698fd05d8c35",
"created_at": 1737578809,
"kind": 1,
"tags": [
[
"t",
"ai"
],
[
"proxy",
"https://tldr.nettime.org/users/tante/statuses/113873964849363495",
"activitypub"
]
],
"content": "Question: Do you know of a paper analyzing whether using a markov chain (as used in many AI crawler tarpits) to generate training data does lead to the same/a smaller/a bigger level of model collapse than generating training data with an LLM? #ai",
"sig": "f48c9144986fc2231ac22259c827c9f12c2d3af8fbe54fe1092505bf8092acacbc1ecec61a8e4d24750739a887668907abbc3c5fbe9eebe8d5b307203f0262dd"
}
