Research map / Vision-language and multimodal / Multimodal LLMs
Fine-Grained Visual Entailment
Christopher Thomas, Yipeng Zhang, Shih-Fu Chang
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 |
|---|---|---|---|
| lower beta_CLS from 1 to 0.5 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_KE->CLS (relabeled): 67.77 -> 80.73 | Improved |
| lower beta_CLS from 1 to 0.5 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_KE: 58.75 -> 68.07 | Improved |
| lower beta_CLS from 1 to 0.5 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_con: 30.85 -> 62.76 | Improved |
| lower beta_CLS further from 0.5 to 0.1 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_con: 62.76 -> 44.68 | Hurt |
| lower beta_CLS further from 0.5 to 0.1 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_KE->CLS (relabeled): 80.73 -> 75.41 | Hurt |
| lower beta_CLS further from 0.5 to 0.1 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_KE: 68.07 -> 68.30 | No clear effect |
| add the structural loss (beta_STRUC 0 -> 1) (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_STRUC: 70.26 -> 96.36 | Improved |
| add the structural loss (beta_STRUC 0 -> 1) (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_con: 9.57 -> 62.76 | Improved |
| add the structural loss (beta_STRUC 0 -> 1) (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_neu: 15.48 -> 31.29 | Improved |
| add the structural loss (beta_STRUC 0 -> 1) (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_ent: 88.87 -> 79.64 | Hurt |
| raise beta_KE from 0.5 to 1 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_KE: 61.68 -> 68.07 | Improved |
| raise beta_KE from 0.5 to 1 (ablation) | Ours (KE classifier) · Fine-grained VE (KE classifier only) | Acc_STRUC: 98.40 -> 96.36 | Hurt |
Closest papers by meaning
- Defeasible Visual Entailment: Benchmark, Evaluator, and Reward-Driven Optimization (AAAI 2025 · similarity 0.87)
- Natural Language Inference Improves Compositionality in Vision-Language Models (ICLR 2025 · similarity 0.83)
- Improved Visual Grounding through Self-Consistent Explanations (CVPR 2024 · similarity 0.82)
- Democratizing Fine-grained Visual Recognition with Large Language Models (ICLR 2024 · similarity 0.82)
- JourneyBench: A Challenging One-Stop Vision-Language Understanding Benchmark of Generated Images (NeurIPS 2024 · similarity 0.82)
- Compositional Entailment Learning for Hyperbolic Vision-Language Models (ICLR 2025 · similarity 0.81)
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.
