Yes, it has the standard LLM trait that when you nudge it to stop being confidently wrong based on insufficient information, it also tends to be less assertive

Yes, it has the standard LLM trait that when you nudge it to stop being confidently wrong based on insufficient information, it also tends to be less assertive when it actually has sufficient information. But I'm not sure why anyone would prefer a model which parses sentences as containing information that isn't there 30-50% of the time to a model which gives false negatives 4-10 %age points more often when given relevant information (especially since the baseline model was already too bad at identifying true positives to be remotely useful at that task)