GPT-3 can sum a list of 10 numbers 22 points by MikaelMayer 4 days ago | hide | past | favorite | 1 comment In * * the prompt with very conservative parameters. This prompts demonstrates that GPT-3 can sum list of numbers, i.e. do iterative reasoning.`````` *GPT-3 can sum lists of up to 10 numbers. Problem: Simulate the execution of 87+18+95+59+13+30+35+25+40+36 Reasoning: Let a=87, b=18, c=95, d=59, e=13, f=30, g=35, h=25, i=40, j=36 a+b=87+18=105=B B+c=105+95=200=C C+d=200+59=259=D D+e=259+13=272=E E+f=272+30=302=F F+g=302+35=337=G G+h=337+25=362=H H+i=362+40=402=I I+j=402+36=438 Solution: 438 Problem: Simulate the execution of 5+74+84+29+59+41+25+1+15+38 Reasoning:* Let a=5, b=74, c=84, d=29, e=59, f=41, g=25, h=1, i=15, j=38 a+b=5+74=79=G G+c=79+84=163=H H+d=163+29=192=I I+e=192+59=251=J J+f=251+41=292=K K+g=292+25=317=L L+h=317+1=318=M M+i=318+15=333=N N+j=333+38=371 Solution: 371 `````` One might wonder, why is all this fluffy stuff with variable names necessary? Here is the story how I came up with the prompt above. I first tried to do the same without these variables, and it fails, and I think I know why.It fails to add the penultimate element (the 9th element of a list of 10 elements), constantly. When I click it, I see its probability was 40% and the probability of last was 60% roughly. It's very similar to how we humans reasons when we are told a list of 10 numbers. We remember the first ones (a sequence) the last one (because being the last is an event), but we might forget about some elements in between, or think we have to finish the computation quickly. I thought, how do we humans augment our memory abilities, like to remember 100 items? From experimenting it myself, I knew that associative memories augment the capacity to remember things. For example, to remember 100 items, I can think of walking through a place of 100 rooms I know, and imagine an item in each room. What would be the equivalent for GPT-3? Associate the numbers to add with something else. I thought, let's put variable names.

 Wow... This is surprising to me.How can it know all these different additions? Has it been trained on so much data that it came across 292+25=317 and is able to summon it in a situation like this? Or did it learn to actually add decimal numbers (the way we do it).Does it always work or what’s the success rate?Did you try with 5-digit numbers and/or negative numbers?

Search: