minus-squareRight_Ad371@alien.topBtoLocalLLaMA•Training LLMs to follow procedure for Math gives an accuracy of 98.5%linkfedilinkEnglisharrow-up1·1 year agoImpressive work. It always bugged me when I had to use the Wolfram plugin to solve simple math problems. Given that GPT 3.5 is being used, is there any chance that GPT 4 would cover up the remaining 1.5% error? linkfedilink
Impressive work. It always bugged me when I had to use the Wolfram plugin to solve simple math problems. Given that GPT 3.5 is being used, is there any chance that GPT 4 would cover up the remaining 1.5% error?