Vitalik Buterin pfp
Vitalik Buterin
@vitalik.eth
A quick test of the new llama3 models (and old ones): L3 70b: https://i.imgur.com/HPgNLnW.png L3 8b: https://i.imgur.com/HTxZmy9.png Mixtral 8x7b: https://i.imgur.com/qjEk93V.png ChatGPT: https://i.imgur.com/DIQ5loP.png ChatGPT is correct, L3 70b almost; others are wrong.
60 replies
105 recasts
668 reactions
trollman๐ŸŽฉ๐Ÿ–๐ŸŒฑโ›“๏ธ๐Ÿ’Ž pfp
trollman๐ŸŽฉ๐Ÿ–๐ŸŒฑโ›“๏ธ๐Ÿ’Ž
@trollman
hello vitalik how do you do?
1 reply
0 recast
1 reaction
Dean Pierce ๐Ÿ‘จโ€๐Ÿ’ป๐ŸŒŽ๐ŸŒ pfp
Dean Pierce ๐Ÿ‘จโ€๐Ÿ’ป๐ŸŒŽ๐ŸŒ
@deanpierce.eth
I feel like the base model of llama2 wasn't great at coding either, but the fine tunes and tweaked architectures made a lot of magic happen. Hopefully we'll see that happen in the coming weeks with llama3.
0 reply
0 recast
2 reactions
Tip pfp
Tip
@tip
@thefreddy, you can use the frame below to send an onchain tip to @vitalik.eth
0 reply
0 recast
0 reaction
Eric Siu ๐Ÿˆ pfp
Eric Siu ๐Ÿˆ
@randomishwalk
whatโ€™s the lmsys leaderboard ELO?
0 reply
0 recast
0 reaction
๐š–_๐š“_๐š› pfp
๐š–_๐š“_๐š›
@m-j-r.eth
the snake game is dead. long live the daylight savings script.
0 reply
0 recast
0 reaction
Ritika ๐ŸŽฉ๐Ÿ–๐Ÿน pfp
Ritika ๐ŸŽฉ๐Ÿ–๐Ÿน
@ritika
Hi Vitalik, any job for reply guys?
0 reply
0 recast
0 reaction
aron pfp
aron
@aron
Iโ€™ve also been playing with these models, itโ€™s a lot about prompt + use case I feel. Like I find Claude opus to be a much better writer compared to chatgpt.
0 reply
0 recast
0 reaction
Yumemaboroshi ๐Ÿซ‚ ๐Ÿ–๐Ÿ”ฎ๐Ÿก pfp
Yumemaboroshi ๐Ÿซ‚ ๐Ÿ–๐Ÿ”ฎ๐Ÿก
@yumemaboroshi
Hey! I've recently purchased your cast! You can now claim your ๐ŸŽฉ DEGEN tokens on jam! ๐ŸŸฃ https://jam.so/ ๐ŸŸฃ
0 reply
0 recast
0 reaction
kidkoala_uk pfp
kidkoala_uk
@kidkoala
Vitalik can you see if it can calculate the length of this.
0 reply
0 recast
2 reactions
cryptobomj pfp
cryptobomj
@cryptobomj1976
ะ’ะธั‚ะฐะปั ัะฟะฐัะธะฑะพ ะทะฐ ะดะตั‚ัั‚ะฒะพ
0 reply
0 recast
1 reaction
fire pepe pfp
fire pepe
@rms7282
Amazing, do you know fire pepe?
0 reply
0 recast
1 reaction
verga pfp
verga
@verga
IO helped $ETH
0 reply
0 recast
0 reaction
Drzee89 pfp
Drzee89
@drzee89
@vitalik.eth have you implemented GROQ api and agents? 7x output, with capability of hierarchical checks along with embeddings to call upon. Enterprise allows fine tuned models to be used with GROQ
0 reply
0 recast
0 reaction
yes, ok ๐Ÿ‘ pfp
yes, ok ๐Ÿ‘
@kiss4867
โ—Attention Warpcast & DEGEN community! Statement on the Recent Mass Disqualifications in the Degen Community.
0 reply
0 recast
0 reaction
Franktheftank pfp
Franktheftank
@franktheftank
Then L3 is very good for local models
0 reply
0 recast
0 reaction
Nila77 pfp
Nila77
@nila77
Mixtral 8x7b: ChatGPT: ChatGPT is correct, L3 70b almost; others are wrong.
0 reply
0 recast
0 reaction
weremainbulish pfp
weremainbulish
@weremainbullish
Incredible
0 reply
0 recast
0 reaction
Cryptodines pfp
Cryptodines
@cryptodines
๐Ÿ‘
0 reply
0 recast
0 reaction
Belladonnaโญ๏ธ๐ŸŒˆ๐Ÿ’ฆ pfp
Belladonnaโญ๏ธ๐ŸŒˆ๐Ÿ’ฆ
@fastflash
the snake game is dead. long live the daylight savings script.
0 reply
0 recast
0 reaction