I have high hopes for Meta’s LLaMA, the 13B model outperforms GPT-3 and should run nicely on a single consumer GPU, even the 65B model which is even more powerful. It will be eventually released under open source license.
https://ai.facebook.com/blog/large-language-model-llama-meta-ai/
streetyogi /
npub1nu…8qmun
2023-03-07 03:48:15
in reply to nevent1q…wcsp
Author Public Key
npub1nujsvauvdlj8rat0rhvrksa56wuwr7k6mteh3pqtfnwnnmtxug9sg8qmunPublished at
2023-03-07 03:48:15Event JSON
{
"id": "264eb304c8ad61ce7f715642373960f1dd0f201612b1e2133157ec6882d81622",
"pubkey": "9f2506778c6fe471f56f1dd83b43b4d3b8e1fadadaf378840b4cdd39ed66e20b",
"created_at": 1678157295,
"kind": 1,
"tags": [
[
"e",
"0b5c27a9bfc22ecf3f7835babe99797d699de2c8b326cb32910c4ab1f5df40ca"
],
[
"e",
"fbd7f1b1b5842db6789e6c7941eb349d3497a7085d51db14279cbdf97e1c781e"
],
[
"p",
"33093204dee6b515e76126e1e587acd732581792c22a20a52cdf5a44d7eeeeff"
]
],
"content": "I have high hopes for Meta’s LLaMA, the 13B model outperforms GPT-3 and should run nicely on a single consumer GPU, even the 65B model which is even more powerful. It will be eventually released under open source license.\n\nhttps://ai.facebook.com/blog/large-language-model-llama-meta-ai/",
"sig": "362f3a8c8fe7e757a3800a2039f812c6b8e39843252405ea8db1924de495ddf0ec95046d7f1b5e014ea52403d42832c1df506cc000f223ab0e7eed9fd22ae7ad"
}