Great article, unpacking Anthropic's findings about how LLMs work, some of…
EN
Great article, unpacking Anthropic's findings about how LLMs work, some of which are very relevant to translation - explaining what users interacting with them had already sensed:
"Even when you input a prompt in Chinese natural language, the model might internally represent that meaning in a form closer to English, especially if its training data is disproportionately English.
Already added to our picks at the AI Localization Think Tank resources page, under "Loc-relevant".
