Prompt caching in #Claude announced today. They claim it reduces costs by “up to 90%”. To me, that says they expect people to use the context instead of #RAG. Include whole databases of text in every prompt.
#LLMs #anthropic
https://www.anthropic.com/news/prompt-caching
Tim Kellogg /
npub1nn…xdwe8
2024-08-14 16:18:58
Author Public Key
npub1nnm8m3ws09epunguu09g7lch8gtup4vn2jnhgtv6zq7xufgc3uesvxdwe8Published at
2024-08-14 16:18:58Event JSON
{
"id": "44503594193137d3a8cade266c31f1f739089da515ad7adbddcc2119cce14f07",
"pubkey": "9cf67dc5d079721e4d1ce3ca8f7f173a17c0d59354a7742d9a103c6e25188f33",
"created_at": 1723652338,
"kind": 1,
"tags": [
[
"t",
"anthropic"
],
[
"t",
"LLMs"
],
[
"t",
"claude"
],
[
"t",
"rag"
],
[
"proxy",
"https://hachyderm.io/@kellogh/112961279686936795",
"web"
],
[
"proxy",
"https://hachyderm.io/users/kellogh/statuses/112961279686936795",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://hachyderm.io/users/kellogh/statuses/112961279686936795",
"pink.momostr"
],
[
"-"
]
],
"content": "Prompt caching in #Claude announced today. They claim it reduces costs by “up to 90%”. To me, that says they expect people to use the context instead of #RAG. Include whole databases of text in every prompt.\n#LLMs #anthropic\nhttps://www.anthropic.com/news/prompt-caching",
"sig": "02f614e45d156229dcce69e3a0589ba52f54bae7499eb73a45087a959e7968f23e30a0578e35ae1f0e2455a93cb385e4a534e1acd063a6ef875cfc5f4a40396b"
}