☆ Yσɠƚԋσʂ ☆ to TechnologyEnglish · 7 months agoBy using the same techniques Google used to solve Go (MTCS and backprop), Llama8B gets 96.7% on math benchmark GSM8K. That’s better than GPT-4, Claude and Gemini, with 200x less parameters!arxiv.orgexternal-linkmessage-square0fedilinkarrow-up118arrow-down12file-textcross-posted to: machinelearning@lemmy.ml
arrow-up116arrow-down1external-linkBy using the same techniques Google used to solve Go (MTCS and backprop), Llama8B gets 96.7% on math benchmark GSM8K. That’s better than GPT-4, Claude and Gemini, with 200x less parameters!arxiv.org☆ Yσɠƚԋσʂ ☆ to TechnologyEnglish · 7 months agomessage-square0fedilinkfile-textcross-posted to: machinelearning@lemmy.ml