KARMA: Knowledge-Action Regularized Multimodal Alignment for Personalized Search at Taobao

1. 1Identify the knowledge-action gap in LLMs for personalized tasks
2. 2Develop a multimodal alignment approach to bridge the gap
3. 3Fine-tune LLMs with knowledge-action regularization for improved performance
4. 4Deploy the KARMA model in a real-world personalized search system

AI engineers and researchers on a team can benefit from KARMA as it improves personalized search results, and product managers can utilize it to enhance user experience

💡 KARMA addresses the suboptimal performance of fine-tuned LLMs in industrial personalized tasks by aligning multimodal knowledge and actions

💡 KARMA bridges the knowledge-action gap in LLMs for personalized search