AI公司Musubi宣布推出轻量级决策模型PolicyLM-1.7B，专门用于实时内容审核1。该模型采用开放权重发布，能够在50毫秒内将纯英文的内容政策应用于消息审核1。与传统审核系统相比，PolicyLM-1.7B无需特殊训练即可应用复杂政策，且政策变更时无需重新训练1。
Musubi联合创始人兼首席AI官Filip Jankovic表示，"能够以非常可扩展、可定制的方式标记所有内容极其有用"1。这项技术在成本和速度上与现有社交平台审核系统相当，但具有现代大语言模型的灵活性1。决策模型领域近期成为AI产业热点，自Typesafe AI的Jev在9月发布以来，包括OpenAI和Amazon在内的主要科技公司相继推出类似竞品1。
AI company Musubi has unveiled PolicyLM-1.7B, a lightweight decision model engineered to streamline real-time content moderation 1. The model can apply English-language content policies to message review within 50 milliseconds without requiring specialized training or model retraining when policies change 1. Released with open weights, PolicyLM-1.7B represents a practical application of decision models—a category of AI systems that has gained prominence following similar releases from OpenAI and Amazon 1.
Musubi's co-founder and chief AI officer Filip Jankovic emphasized the utility of the approach, noting that "being able to flag all content in a very scalable, customizable way is extremely useful" 1. The model delivers performance comparable to existing social platform moderation systems in terms of cost and speed, while incorporating the flexibility of modern large language models 1. Decision models have become a focal point in AI development since their introduction by Typesafe AI's Jev in September 1.
评论
还没有评论，欢迎留下第一条。