Comment by vatsachak

Why is ML so misleading these days? No this model does not score 80%+ on DeepSWE, it merely chooses the best possible idea of Opus 5 at every stage, increasing the performance by 5%.