Event JSON
{
"id": "97d6d6358d050870bd40f2daefaaf2f101285684eb16b7f6a0283e32064d4bf1",
"pubkey": "ece51acfc6c84a5f625a755bed921f6eac6a9daa6f90817047901dbe0231522f",
"created_at": 1731244371,
"kind": 1,
"tags": [
[
"p",
"2b4287d5a1e17396428bf537857b91c2355e6ce3f0800dc82ee22a64521b0a0b",
"wss://relay.mostr.pub"
],
[
"p",
"ebdd255d38b43576b4a85e0f64879ef80d7bcc6dd2713236091ed8e27a5505cd",
"wss://relay.mostr.pub"
],
[
"e",
"cf7de5f661d622f760484ff44d4c63ca2a49c3588e2d4bb2da663567b99ed07f",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://mastodon.social/users/mistersql/statuses/113458831153554017",
"activitypub"
]
],
"content": "nostr:npub19dpg04dpu9eevs5t75mc27u3cg64um8r7zqqmjpwug4xg5smpg9s7lkv2q Matches my experience, too. I think on current consumer hardware the 7b models are less for asking questions about JS \u0026 more for the gap between things ordinary code can't do but LLMs find easy even with a 7b model, things like Named Entity Extraction, parsing dates out of a totally arbitrary pattern. Anything that needs to be like a skilled software developer you need a frontier model like 4o or Claude.",
"sig": "62309d6a2a18c3b4793eb61d8e7ad508189d956b2b0ef32cd9140a52af4161b244d1f08b9b13aa5edf9aa90b3fde6832e2e95031812611dcb1469c845d938843"
}