7 papers
Beyond Sparse Rewards: A New Benchmark and Structure-Aware Graph Alignment for Micro-Drama Understanding
Yixin Qin, Shi-Zhe Chen, Zhiqi Yu +4
Micro-dramas, characterized by ultra-short durations and hyper-dense storylines, pose unique challenges for video understanding that conventional benchmarks fail to address. To bri…
FIRE-CIR: Fine-grained Reasoning for Composed Fashion Image Retrieval
François Gardères, Camille-Sovanneary Gauthier, Jean Ponce +1
Composed image retrieval (CIR) aims to retrieve a target image that depicts a reference image modified by a textual description. While recent vision-language models (VLMs) achieve…
HyLaR: Hybrid Latent Reasoning with Decoupled Policy Optimization
Tao Cheng, Shi-Zhe Chen, Hao Zhang +3
Chain-of-Thought (CoT) reasoning significantly elevates the complex problem-solving capabilities of multimodal large language models (MLLMs). However, adapting CoT to vision typica…
Hear: Hierarchically Enhanced Aesthetic Representations For Multidimensional Music Evaluation
Shuyang Liu, Yuan Jin, Rui Lin +3
Evaluating song aesthetics is challenging due to the multidimensional nature of musical perception and the scarcity of labeled data. We propose HEAR, a robust music aesthetic evalu…
Conan-Embedding-v2: Training an LLM from Scratch for Text Embeddings
Shiyu Li, Yang Tang, Ruijie Liu +2
Large language models (LLMs) have recently demonstrated excellent performance in text embedding tasks. Previous work usually use LoRA to fine-tune existing LLMs, which are limited…
U-MARVEL: Unveiling Key Factors for Universal Multimodal Retrieval via Embedding Learning with MLLMs
Xiaojie Li, Chu Li, Shi-Zhe Chen +1
Universal multimodal retrieval (UMR), which aims to address complex retrieval tasks where both queries and candidates span diverse modalities, has been significantly advanced by th…
Conan-embedding: General Text Embedding with More and Better Negative Samples
Shiyu Li, Yang Tang, Shizhe Chen +1
With the growing popularity of RAG, the capabilities of embedding models are gaining increasing attention. Embedding models are primarily trained through contrastive loss learning,…