nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqjpp36eeycvlkkdy4d4l4lgswvf8jtdare5vly0w0sk7rc39pzgrqr9dhpg (nprofile…dhpg) nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqyhj4ua6580h3w0kucwwrtwl4uv9pg7m2drxzplq5x3364jczp36sfkq54u (nprofile…q54u) nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqy4nygj85ggturx4nlzzmdqunaxzsallu2cltm50f37kunx0h7jcsgemhl7 (nprofile…mhl7) nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqn702fwc8fhzsuff296qxsywjpxw2g84x5j7dahk9utjgthage7pq3t3zdw (nprofile…3zdw) nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpq0xnrwauhxqr6q4e3a32exkl3j9z290qs9r85ce4ce7yeg7xup28slw0af3 (nprofile…0af3)
Perhaps we’re getting too philosophical 😉. As a dev, I could agree that the “source code” is the model.
But if you cannot run the source code (without the weights, without the proprietary training data) & get the same final result (that consumers are paying for), then consumers are getting a combined product, and they/others currently call *that* the Model.
If weights are compiled input data, input data needs to be equally open source… right?
Frederic Barthelemy /
npub1cd…2umrq
2025-01-05 00:59:54
in reply to nevent1q…fhlz
Author Public Key
npub1cdrk9puzk3msn8s33zqdew4s0swxyp4hjaah53349fwet7m9vm9qn2umrqPublished at
2025-01-05 00:59:54Event JSON
{
"id": "cc5d2759dac230ddbadb2be2e826f4bac0509a2da5f0b93aa96599c824b60a47",
"pubkey": "c347628782b477099e118880dcbab07c1c6206b7977b7a46352a5d95fb6566ca",
"created_at": 1736038794,
"kind": 1,
"tags": [
[
"p",
"90431d6724c33f6b34956d7f5fa20e624f25b7a3cd19f23dcf85bc3c44a11206",
"wss://relay.mostr.pub"
],
[
"p",
"25e55e77543bef173edcc39c35bbf5e30a147b6a68cc20fc143463aacb020c75",
"wss://relay.mostr.pub"
],
[
"p",
"25664448f44217c19ab3f885b68393e9850efffc563ebdd1e98fadc999f7f4b1",
"wss://relay.mostr.pub"
],
[
"p",
"9f9ea4bb074dc50e252a2e806811d2099ca41ea6a4bcdedec5e2e485dfa8cf82",
"wss://relay.mostr.pub"
],
[
"p",
"79a63777973007a05731ec55935bf19144a2bc1028cf4c66b8cf899478dc0a8f",
"wss://relay.mostr.pub"
],
[
"p",
"e653eb60b2db5edf906d354c2825bf00a80dfab1b1766cd97805155e8fcc0db3",
"wss://relay.mostr.pub"
],
[
"p",
"6e5bb3f4df765d024066fb8b1b0ab196b68726b8092782a387c9698fd05d8c35",
"wss://relay.mostr.pub"
],
[
"p",
"80b0194e84aeff7f305922eeeecf9d5902ed3da2958959139f23661169fa9ac3",
"wss://relay.mostr.pub"
],
[
"p",
"9a319c452c8cf90d2ce1ffb5121bbe020a2fa05d907249dd702c60538aa62607",
"wss://relay.mostr.pub"
],
[
"p",
"6b6e1dee1b859443725fed5d364ece7237ddfe3e6f975c755b8a4bc52dcb44f9",
"wss://relay.mostr.pub"
],
[
"e",
"28c530bcdc8f2c47d580d65941f2c25ec83414c9759b89d5c895f25df65d54f9",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://mastodon.social/users/fbartho/statuses/113773038453672658",
"activitypub"
]
],
"content": "nostr:nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqjpp36eeycvlkkdy4d4l4lgswvf8jtdare5vly0w0sk7rc39pzgrqr9dhpg nostr:nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqyhj4ua6580h3w0kucwwrtwl4uv9pg7m2drxzplq5x3364jczp36sfkq54u nostr:nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqy4nygj85ggturx4nlzzmdqunaxzsallu2cltm50f37kunx0h7jcsgemhl7 nostr:nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqn702fwc8fhzsuff296qxsywjpxw2g84x5j7dahk9utjgthage7pq3t3zdw nostr:nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpq0xnrwauhxqr6q4e3a32exkl3j9z290qs9r85ce4ce7yeg7xup28slw0af3 \n\nPerhaps we’re getting too philosophical 😉. As a dev, I could agree that the “source code” is the model.\n\nBut if you cannot run the source code (without the weights, without the proprietary training data) \u0026 get the same final result (that consumers are paying for), then consumers are getting a combined product, and they/others currently call *that* the Model.\n\nIf weights are compiled input data, input data needs to be equally open source… right?",
"sig": "8ab4a90a3aa3aa70c166e1792aba0c94e32d659b390e77facad0375779705d8c8e215228e70886585993dc483152e2a568f31857b0e3cd427ecc8859fb7ba3ed"
}