Or, if we consider the fact that an LLM’s performance depends on the task’s similarity to others in the training set, it could be that one person is doing a fairly novel task and another is doing something very well represented in online code.
Or, if we consider the fact that an LLM’s performance depends on the task’s similarity to others in the training set, it could be that one person is doing a fairly novel task and another is doing something very well represented in online code.