jsm on Nostr: Maybe, but the S1 paper showed that to really achieve top level performance you ...
Maybe, but the S1 paper showed that to really achieve top level performance you really only need an extremely small dataset of high quality reasoning examples. That’s enough for a normal LLM to pick up on the patterns and you’re good to go.
Published at
2025-02-11 06:08:01Event JSON
{
"id": "d92ab7f343f0af9ccd5ad8074f382a4ec865bb504df3893d4d4a0fd583a2eb2a",
"pubkey": "e0339348ca6cac9708cd98e631e2f4baad534dfce870881b65aa57d30ff7253e",
"created_at": 1739254081,
"kind": 1,
"tags": [
[
"e",
"4eb53815a86c848e5c1d9b0a44f2a809eaf9227f58f8abfb8778a1b5cee821e8",
"ws://192.168.18.7:7777",
"root"
],
[
"p",
"32e1827635450ebb3c5a7d12c1f8e7b2b514439ac10a67eef3d9fd9c5c68e245"
]
],
"content": "Maybe, but the S1 paper showed that to really achieve top level performance you really only need an extremely small dataset of high quality reasoning examples. That’s enough for a normal LLM to pick up on the patterns and you’re good to go.",
"sig": "c4369e7d9683ae9941b1cee2b968e05ecc3c2af8e5a9e52e6b48e212e806385e8c423de93a97cff1e8280603af5c8e0be9b6c401fdd1553970ffc190a01746da"
}
