RedOne: Revealing Domain-specific LLM Post-Training in Social Networking Services

2025-07-12

AI Summary

Existing large language models (LLMs) exhibit limited generalization and domain adaptability for social networking services (SNS), primarily due to their reliance on single-task paradigms, hindering effective content moderation and interaction quality enhancement. Method: We propose RedOne—the first LLM specifically designed for SNS—employing a three-stage training strategy: continual pretraining, supervised fine-tuning, and preference optimization, all grounded in large-scale real-world SNS data to achieve robust domain adaptation. RedOne features a unified foundational architecture supporting eight core SNS tasks, balancing general linguistic competence with domain-specific expertise. Results: Experiments demonstrate an average 14.02% improvement across eight SNS tasks and a 7.56% gain on a bilingual SNS evaluation benchmark. Online deployment shows a 11.23% reduction in harmful content exposure and a 14.95% increase in search-page click-through rate, validating the efficacy and practicality of domain-specialized modeling.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Application Domains: Social Networks

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

As a primary medium for modern information dissemination, social networking services (SNS) have experienced rapid growth, which has proposed significant challenges for platform content management and interaction quality improvement. Recently, the development of large language models (LLMs) has offered potential solutions but existing studies focus on isolated tasks, which not only encounter diminishing benefit from the data scaling within individual scenarios but also fail to flexibly adapt to diverse real-world context. To address these challenges, we introduce RedOne, a domain-specific LLM designed to break the performance bottleneck of single-task baselines and establish a comprehensive foundation for the SNS. RedOne was developed through a three-stage training strategy consisting of continue pretraining, supervised fine-tuning, and preference optimization, using a large-scale real-world dataset. Through extensive experiments, RedOne maintains strong general capabilities, and achieves an average improvement up to 14.02% across 8 major SNS tasks and 7.56% in SNS bilingual evaluation benchmark, compared with base models. Furthermore, through online testing, RedOne reduced the exposure rate in harmful content detection by 11.23% and improved the click page rate in post-view search by 14.95% compared with single-tasks finetuned baseline models. These results establish RedOne as a robust domain-specific LLM for SNS, demonstrating excellent generalization across various tasks and promising applicability in real-world scenarios.

Problem

• Enhancing content management in social networking services
• Overcoming limitations of single-task LLM approaches
• Improving multilingual performance in SNS applications

Innovation

• Three-stage training strategy for LLM
• Domain-specific LLM for SNS tasks
• Large-scale real-world dataset utilization