Comment by mountainriver

LLMs aren't good at deep ML problems yet. We know this, its what MLE-bench is for. That doesn't mean they aren't good at other coding problems