Introducing Code Llama, a state-of-the-art large language model for codingai.meta.comexternal-linkcross-posted to: eric_posts_urls@discuss.onlineaistuff@lemdro.idprogramming@programming.devtechnews@radiation.party ijeff ( @ijeff@lemdro.id ) TechnologyEnglish • 1 year ago message-square2fedilinkarrow-up144
arrow-up144external-linkIntroducing Code Llama, a state-of-the-art large language model for codingai.meta.com ijeff ( @ijeff@lemdro.id ) TechnologyEnglish • 1 year ago message-square2fedilinkcross-posted to: eric_posts_urls@discuss.onlineaistuff@lemdro.idprogramming@programming.devtechnews@radiation.party
minus-square d3Xt3r ( @d3Xt3r@beehaw.org ) linkfedilink12•edit-21 year agoLooks interesting, but doesn’t seem better than GPT-4. GPT-4 scored 67% on the Human Eval test, whereas Code Llama scored only a 53.7%, which isn’t a trivial difference. Bit disingenuous of Meta to claim it to be “on par” with ChatGPT.
minus-square ijeff ( @ijeff@lemdro.id ) OPlinkfedilinkEnglish6•1 year agoThey seem to qualify a bit below that they mean GPT-3.5-Turbo, which does often get referred to as ChatGPT (in contrast to GPT-4).
Looks interesting, but doesn’t seem better than GPT-4. GPT-4 scored 67% on the Human Eval test, whereas Code Llama scored only a 53.7%, which isn’t a trivial difference. Bit disingenuous of Meta to claim it to be “on par” with ChatGPT.
They seem to qualify a bit below that they mean GPT-3.5-Turbo, which does often get referred to as ChatGPT (in contrast to GPT-4).