Research map / Vision-language and multimodal / CLIP and zero-shot
Teach CLIP to Develop a Number Sense for Ordinal Regression
Yao DU, Qiang Zhai, Weihang Dai, Xiaomeng Li
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 ranking-based feature regularizer (FCRC) to coarse-to-fine baseline (ablation) | NumCLIP baseline · MORPH II | MAE: 2.25 -> 2.15 | Improved |
| add linguistic concept mapping of numbers to coarse-to-fine baseline (ablation) | NumCLIP baseline · MORPH II | MAE: 2.25 -> 2.16 | Improved |
| combine linguistic concepts with ranking regularizer (ablation) | NumCLIP baseline · MORPH II | MAE: 2.25 -> 2.08 | Improved |
| use coarse-to-fine paradigm alone versus OrdinalCLIP (ablation) | CLIP ordinal regression · MORPH II | MAE: 2.32 -> 2.25 | Improved |
| use squared or square-root label distance instead of absolute distance for negative weights (ablation) | NumCLIP · MORPH II | MAE: 2.08 -> 2.11 / 2.12 | No clear effect |
| fine-tune CLIP prompts (CoOp) instead of zero-shot | CLIP · MORPH II | MAE: 6.09 -> 2.39 | Improved |
| NumCLIP versus L2RCLIP in few-shot (1-shot) | CLIP ordinal regression · MORPH II 1-shot | MAE: 4.54 -> 3.61 | Improved |
| NumCLIP versus L2RCLIP at 32-shot | CLIP ordinal regression · MORPH II 32-shot | MAE: 2.55 -> 2.60 | Hurt |
| NumCLIP versus L2RCLIP on historical image dating | CLIP ordinal regression · Historical Image Dating | accuracy (%): 67.22 -> 69.61 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OrdinalCLIP — OrdinalCLIP: Learning Rank Prompts for Language-Guided Ordinal Regression (NeurIPS 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Ranking-aware adapter for text-driven image ordering with CLIP (ICLR 2025 · similarity 0.88)
- Demystifying CLIP Data (ICLR 2024 · similarity 0.87)
- TripletCLIP: Improving Compositional Reasoning of CLIP via Synthetic Vision-Language Negatives (NeurIPS 2024 · similarity 0.87)
- Diffusion Feedback Helps CLIP See Better (ICLR 2025 · similarity 0.86)
- Post-pre-training for Modality Alignment in Vision-Language Foundation Models (CVPR 2025 · similarity 0.86)
- Calibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotations (CVPR 2024 · 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.
