I was wrong, You CANNOT make the evaluator smarter/ into AI, what I said about the set of predictions "street 80%, road 10%, hall 10%" VS "street 80%, car 10%, wall 10%" for being "better" for "I was walking down the" only improves the predictions, not the evaluator. Hall makes street light up more, you need to look at all your top predictions. If you are "done" making your predictions and is the final set "street 80%, road 10%, hall 10%", you cannot improves the prediction then any further. The evaluator only stores which word/letter/bit you want it to, it takes 1 bit to store a bit if your predictions are neutrally good, that's all the evaluator does, the predictor is what makes compression of data better.
Sorry I missed this, but at least I caught myself all on my own ~3 months later!! ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Te02b4078ca0cde00-M71d89999c165d5127cedd275 Delivery options: https://agi.topicbox.com/groups/agi/subscription
