Research map / Vision-language and multimodal / CLIP and zero-shot
Conditional Prompt Learning for Vision-Language Models
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 image-conditioned prompt tokens to static learned context | CoOp / CLIP ViT-B/16 · 11-dataset average, new classes | accuracy: 63.22 -> 71.69 | Helped |
| Add image-conditioned prompt tokens to static learned context | CoOp / CLIP ViT-B/16 · 11-dataset average, base classes | accuracy: 82.69 -> 80.47 | Hurt |
| Add image-conditioned prompt tokens to static learned context | CoOp / CLIP ViT-B/16 · 11-dataset average, base and new classes | H: 71.66 -> 75.83 | Helped |
| Transfer image-conditioned rather than static learned prompts | CoOp / CLIP ViT-B/16 · 10-target cross-dataset transfer average | accuracy: 63.88 -> 65.74 | Helped |
| Use image-conditioned rather than static prompts | CoOp / CLIP ViT-B/16 · ImageNet domain generalization targets | accuracy: ImageNetV2: 64.20 -> 64.07; ImageNet-Sketch: 47.99 -> 48.75; ImageNet-A: 49.71 -> 50.63; ImageNet-R: 75.21 -> 76.18 | Mixed |
| Increase static prompt context from 4 to 60 tokens instead of adding Meta-Net (ablation) | CoOp / CLIP ViT-B/16 · ImageNet, new classes | accuracy: 67.88 -> 65.34 | Hurt |
| Add Meta-Net to 4 context tokens rather than simply increasing static context length (ablation) | CoOp / CLIP ViT-B/16 · ImageNet, new classes | accuracy: 67.88 -> 70.43 | Helped |
| Initialize context vectors with word embeddings rather than random Gaussian vectors (ablation) | CoCoOp · 11-dataset base-to-new generalization | accuracy | Helped |
| Use longer context length (8 or 16 tokens) rather than 4 tokens (ablation) | CoCoOp · 11-dataset base-to-new generalization, new classes | accuracy | Helped |
Closest papers by meaning
- Learning to Learn Better Visual Prompts (AAAI 2024 · similarity 0.92)
- IPO: Interpretable Prompt Optimization for Vision-Language Models (NeurIPS 2024 · similarity 0.90)
- CoPL: Contextual Prompt Learning for Vision-Language Understanding (AAAI 2024 · similarity 0.90)
- Learning to Prompt with Text Only Supervision for Vision-Language Models (AAAI 2025 · similarity 0.90)
- Concept-Guided Prompt Learning for Generalization in Vision-Language Models (AAAI 2024 · similarity 0.89)
- Make Prompts Adaptable: Bayesian Modeling for Vision-Language Prompt Learning with Data-Dependent Prior (AAAI 2024 · 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.