OneEmo: A Unified Multimodal Reasoning Model for Emotion Perception

Jiahao Huang · hf · 2026-08-10

Researchers introduced OneEmo, a unified affective generalist model capable of mastering emotion perception, comprehension, and interaction.

• New Dataset: Constructed EmoWorld-130K, a comprehensive dataset that distills specialized affective knowledge into explicit reasoning trajectories via a human-in-the-loop workflow.
• New Strategy: Proposed Emo-Chord, a novel reinforcement learning strategy that stabilizes optimization through unified multi-task reward allocation to unlock latent reasoning potential.
• Performance: Achieves state-of-the-art performance across most benchmarks. Despite having significantly fewer parameters than commercial models, it delivers highly competitive results.

Original post →

More from Research

• Offline KD Boosts Throughput 41% on Single H200, Slashes LLM Distillation Memory — MultiverseComputingCAI · 2026-08-10
• FATE Model: Achieving Dual Frame-Level Semantic and Temporal Alignment for Audio-Visual — RUC · 2026-08-10
• Hardcore Reverse Engineering: Developer Rebuilds Kimi K3 Training Pipeline from Scratch — sharpeye_wnl · 2026-08-10
• A New Approach to Inference Costs: Exploring Server-Edge Split Model Architectures — komorra · 2026-08-10
• Leaked Architecture of ~400B MoE Model with Aggressive GQA Sparks Interest — teortaxesTex · 2026-08-10
• Research: Unconditional Prediction Accuracy Isn't Always the Right Objective in AI Decision Processes — joshgans · 2026-08-10

Research channel →