tomhazledine on Nostr: Starting to feel pretty pleased with the #LLM generated “related posts" section at ...
Starting to feel pretty pleased with the #LLM generated “related posts" section at the bottom of all my blog posts.
Similarity is calculated by using embeddings and cosine-similarity, and the description is generated using #openai #GPT4
Tweaked the prompt to make the recommendations a little terser and more fact-based, which has made a big difference.
(Massively inspired by listening to
npub1f6a33pfyp67y8llhunlhrf855xm47n3fdqymvxfj7yx78c6vqf4scxpnql (npub1f6a…pnql) talk about the power of embeddings - and you know what? I think he was right!)
Published at
2023-10-20 16:12:04Event JSON
{
"id": "10eaa0e40f1a0e0157421cbbf021d54d038fae467c7983fe7ea45017ec296046",
"pubkey": "8b387cc18c92a0b4d42bdc431c93d10c3e2ee06eda3c869760e44af4174b7495",
"created_at": 1697818324,
"kind": 1,
"tags": [
[
"p",
"4ebb1885240ebc43fff7e4ff71a4f4a1b75f4e296809b61932f10de3e34c026b",
"wss://relay.mostr.pub"
],
[
"p",
"8b0be93ed69c30e9a68159fd384fd8308ce4bbf16c39e840e0803dcb6c08720e",
"wss://relay.mostr.pub"
],
[
"t",
"llm"
],
[
"t",
"openai"
],
[
"t",
"GPT4"
],
[
"proxy",
"https://mastodon.social/users/tomhazledine/statuses/111268221711301875",
"activitypub"
]
],
"content": "Starting to feel pretty pleased with the #LLM generated “related posts\" section at the bottom of all my blog posts.\n\nSimilarity is calculated by using embeddings and cosine-similarity, and the description is generated using #openai #GPT4\n\nTweaked the prompt to make the recommendations a little terser and more fact-based, which has made a big difference.\n\n(Massively inspired by listening to nostr:npub1f6a33pfyp67y8llhunlhrf855xm47n3fdqymvxfj7yx78c6vqf4scxpnql talk about the power of embeddings - and you know what? I think he was right!)\n\nhttps://files.mastodon.social/media_attachments/files/111/268/221/423/793/396/original/7e07f8db706ecd23.png",
"sig": "78e35b34b10ea148f1d502e9940161a013d9bf16fb82b79d2955a0c575fa0004f87a6fabbe462757c046728d9b9a204439189e24d4bca80ab304f9121d3c5083"
}
