Alphacode 2 technical paper claims to solve 43% of problems (77 problems from 12 codeforce competitions) performing 85%ile on all human participants. Caveat is

Alphacode 2 technical paper claims to solve 43% of problems (77 problems from 12 codeforce competitions) performing 85%ile on all human participants. Caveat is deep in the technical paper, 1) generates 1m candidates from N different prior models (fine tuned on previous codeforces) 2) throws away 95%ile candidates (doesn't fit the test case + no compile) 3) groups semantically similar candidates 4) scores candidates from each group (Based on another scoring model, probably latency + descriptions etc) 5) Picks top 10 Makes 10 submissions and finally gets the score .. sure this is how humans solve problems ... totally awed by AGI /s