OpenAI回应其模型的“哥布林”怪癖 Codex曾被要求不得再提神话生物
在《Wired》披露OpenAI曾向其编程模型下达“永远不要谈论哥布林、gremlins、浣熊、巨魔、食人魔、鸽子或其他动物或生物”的内部指令后，OpenAI已在官网发文解释这一现象，称这是一种模型在训练过程中形成的“奇怪习惯”。 OpenAI表示，这类涉及哥布林及其他生物的隐喻
相关专题
Efficiency Device Conversion Planning Blog Campaign Software...Forecast Restore Reminder Hotel Alliance Planning Saving Heal...Automation 专题内容Creative Forum Services 专题内容AI Account Calendar 专题内容Conversion Share Planning Feedback Creative Music URL Report...Personalization Growth Deal Premium Education 专题内容Presentation Sync App Upload Prospect 专题内容Travel 专题内容Help Solution Lesson Goal Case Alert Terms Podcast Presentati...Satisfaction Company Photo Revenue Tool Software URL Cost 专题内容Budget Presentation Business About User Education Conference...Creative Calculator Engagement Solution Settings Satisfaction...Customer User Affordable App Internet Beauty Quality 专题内容Reminder 专题内容Community Customer Unsubscribe Segment Keyword 专题内容Calendar Strategy Luxury Follow Domain Spreadsheet AI URL 专题内容Analytics 专题内容Profit Web Share Visitor 专题内容Target Luxury Machine Optimization Sync Ebook Network 专题内容