jsm on Nostr: It’s crazy to see how much my goalposts have changed for LLMs. I remember being ...
It’s crazy to see how much my goalposts have changed for LLMs. I remember being blown away that GPT-3 could sensibly continue text while using context clues from MULTIPLE SENTENCES AGO. Now I’m annoyed when the 1000 lines of syntactically correct code Claude 3.5 Sonnet generates to satisfy my vague question contains a bug.
Published at
2024-09-01 01:25:04Event JSON
{
"id": "a4f242618c5429fc8041cf28bb6bc64012374196b0178ac35f4723b868fb0f81",
"pubkey": "e0339348ca6cac9708cd98e631e2f4baad534dfce870881b65aa57d30ff7253e",
"created_at": 1725153904,
"kind": 1,
"tags": [],
"content": "It’s crazy to see how much my goalposts have changed for LLMs. I remember being blown away that GPT-3 could sensibly continue text while using context clues from MULTIPLE SENTENCES AGO. Now I’m annoyed when the 1000 lines of syntactically correct code Claude 3.5 Sonnet generates to satisfy my vague question contains a bug.",
"sig": "1e604eb882cb558c9a3b80584c72b6a48dee990c66361c28ceef2291e38ef042bc015b5d4dc9efe25fa55055df81ffdfe8eedce5cbcbe5f31a7344450d0154af"
}
