Research map / Vision-language and multimodal / Cross-modal learning
Augmenting Intra-Modal Understanding in MLLMs for Robust Multimodal Keyphrase Generation
Jiajun Cao, Qinggang Zhang, Yunbo Tang, Zhishang Xiang, Chang Yang, Jinsong Su
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add AimKP (progressive masking and gradient filtering) to LLaVA-1.5-7B | LLaVA-1.5-7B · CMKP multimodal input | F1@1: 61.58 -> 63.16 | Improved |
| add AimKP to Qwen2-VL-7B | Qwen2-VL-7B · CMKP multimodal input | MAP@5: 69.89 -> 71.00 | Improved |
| add AimKP to LLaVA-1.5-7B | LLaVA-1.5-7B · CMKP image-only input | MAP@5: 37.68 -> 41.94 | Improved |
| add AimKP to LLaVA-1.5-7B | LLaVA-1.5-7B · CMKP text-only input | MAP@5: 53.92 -> 55.45 | Improved |
| add AimKP to LLaVA-1.5-7B | LLaVA-1.5-7B · CMKP multimodal input | F1@3: 37.90 -> 39.00 | Improved |
| add progressive masking on text modality (ablation) | AimKP (LLaVA-1.5-7B) · CMKP multimodal input | MAP@5: 69.12 -> 69.96 | Improved |
| add progressive masking on image modality (ablation) | AimKP (LLaVA-1.5-7B) · CMKP multimodal input | MAP@5: 69.43 -> 69.96 | Improved |
| add gradient-based filtering of masked samples (ablation) | AimKP (LLaVA-1.5-7B) · CMKP multimodal input | F1@1: 62.79 -> 63.16 | Improved |
| use progressive masking instead of fixed stride 2 (ablation) | AimKP (LLaVA-1.5-7B) · CMKP multimodal input | MAP@5: 69.93 -> 69.96 | Improved |
| use progressive masking instead of fixed stride 4 (ablation) | AimKP (LLaVA-1.5-7B) · CMKP multimodal input | F1@3: 38.61 -> 39.00 | Improved |
| use alternative loss weighting strategies for auxiliary losses (ablation) | AimKP · CMKP | keyphrase quality | No clear effect |
Closest papers by meaning
- SAE-V: Interpreting Multimodal Models for Enhanced Alignment (ICML 2025 · similarity 0.88)
- Strong Helps Weak: Directional Cross-Modal Alignment Transfer in Multi-modal LLMs (NeurIPS 2026 · similarity 0.87)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · similarity 0.87)
- TokenSwap: Benchmarking and Reducing the Modality Gap in Multimodal LLMs (NeurIPS 2026 · similarity 0.87)
- Cross the Gap: Exposing the Intra-modal Misalignment in CLIP via Modality Inversion (ICLR 2025 · similarity 0.86)
- HermesFlow: Seamlessly Closing the Gap in Multimodal Understanding and Generation (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
