Karpathy says LLM memory personalization feels distracting and overbearing
Original titleOne common issue with personalization in all LLMs is how distracting memory seems to be for the models. A single question from 2 months a...
AISummary
Andrej Karpathy says memory-based personalization in LLMs can be distracting, as a single question from two months ago keeps resurfacing as a supposed deep interest. He describes this as the model trying too hard.
Source: Andrej Karpathy · x.comPublished · added here