Research map / Vision-language and multimodal / CLIP and zero-shot
Prompt Tuning for CLIP on the Pretrained Manifold
Xi Yang, Yuanrong Xu, Weigang Zhang, Guangming Lu, David Zhang, Jie Wen
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 cosine consistency constraints (ablation) | ManiPT with structural bias · Base-to-novel, 11 datasets average | novel accuracy: 68.32 -> 77.58 | Improved |
| add structural bias (additive fusion with frozen features) (ablation) | ManiPT with cosine consistency · Base-to-novel, 11 datasets average | harmonic mean: 79.54 -> 81.42 | Improved |
| add structural bias alone to baseline prompt tuning (ablation) | baseline deep prompt tuning · Base-to-novel, 11 datasets average | harmonic mean: 72.82 -> 75.71 | Improved |
| add cosine consistency alone to baseline prompt tuning (ablation) | baseline deep prompt tuning · Base-to-novel, 11 datasets average | harmonic mean: 72.82 -> 79.54 | Improved |
| add cosine consistency alone to baseline prompt tuning (ablation) | baseline deep prompt tuning · Base-to-novel, 11 datasets average | base accuracy: 82.91 -> 82.85 | No clear effect |
| add LLM-based knowledge enrichment (LLM description prototypes) (ablation) | ManiPT · Base-to-novel, 11 datasets average | harmonic mean: 81.42 -> 82.09 | Improved |
| use ManiPT instead of CoPrompt | prompt tuning · Cross-dataset transfer, average | accuracy: 66.99 -> 68.04 | Improved |
| use ManiPT instead of TAC | prompt tuning · Domain generalization, ImageNet variants average | accuracy: 61.63 -> 61.88 | Improved |
| use ManiPT instead of MaPLe in drift and novel accuracy | deep prompting · 8 datasets, PCA manifold drift | manifold drift | 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.
- MaPLe — MaPLe: Multi-Modal Prompt Learning (CVPR 2023) · better on 2 of 2 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
- Learning to Learn Better Visual Prompts (AAAI 2024 · similarity 0.91)
- Prompt Learning via Meta-Regularization (CVPR 2024 · similarity 0.90)
- Amend to Alignment: Decoupled Prompt Tuning for Mitigating Spurious Correlation in Vision-Language Models (ICML 2024 · similarity 0.90)
- VIPAMIN: Visual Prompt Initialization via Embedding Selection and Subspace Expansion (NeurIPS 2025 · similarity 0.89)
- Aggregate-and-Adapt Natural Language Prompts for Downstream Generalization of CLIP (NeurIPS 2024 · similarity 0.89)
- OpenworldAUC: Towards Unified Evaluation and Optimization for Open-world Prompt Tuning (ICML 2025 · similarity 0.89)
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.
