Research map / Vision-language and multimodal / CLIP and zero-shot
Quantized Prompt for Efficient Generalization of Vision-Language Models
Tianxiang Hao, Xiaohan Ding, Juexiao Feng, Yuhong Yang, Hui Chen, Guiguang Ding
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 moderate Gaussian noise to prompts (not too small or large) (ablation) | prompt tuning (CoOp) · Base-to-new | new-class accuracy | Improved |
| quantize prompts with K-Means QAT (QCoOp) instead of fp16 CoOp | CoOp · Base-to-new (11 datasets avg) | harmonic mean: 71.66 -> 77.43 | Improved |
| quantize MaPLe prompts and linear weights (QMaPLe) | MaPLe · Base-to-new (11 datasets avg) | harmonic mean: 78.55 -> 79.12 | Improved |
| add normalization/denormalization before K-Means quantization (ablation) | QCoOp · Base-to-new | harmonic mean: 75.50 -> 75.85 | Improved |
| add constrained adaptive clustering (ablation) | QCoOp · Base-to-new | harmonic mean: 75.85 -> 76.07 | Improved |
| use QAT instead of PTQ for 1-bit quantization (ablation) | CoOp prompts · Base-to-new | harmonic mean: 74.73 -> 76.31 | Improved |
| increase quantization bits from 1 to 4 (ablation) | QCoOp · Base-to-new | harmonic mean: 75.92 -> 75.76 | No clear effect |
| add prompts in more transformer layers (depth 1 to 6) (ablation) | QCoOp · Base-to-new | harmonic mean: 75.92 -> 76.90 | Mixed |
| quantize visual prompts instead of text prompts (ablation) | QCoOp · Base-to-new | harmonic mean: 75.92 -> 70.98 | Hurt |
| apply QCoOp to SLIP backbone instead of CoOp | SLIP · Base-to-new (SLIP) | harmonic mean: 55.51 -> 71.07 | Improved |
| QCoOp vs CoOp on domain generalization | QCoOp · Domain generalization | target average accuracy: 59.28 -> 60.20 | 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
- IPO: Interpretable Prompt Optimization for Vision-Language Models (NeurIPS 2024 · similarity 0.90)
- Prompt Learning via Meta-Regularization (CVPR 2024 · similarity 0.90)
- ZIP: An Efficient Zeroth-order Prompt Tuning for Black-box Vision-Language Models (ICLR 2025 · similarity 0.90)
- Aggregate-and-Adapt Natural Language Prompts for Downstream Generalization of CLIP (NeurIPS 2024 · similarity 0.90)
- On the Test-Time Zero-Shot Generalization of Vision-Language Models: Do We Really Need Prompt Learning? (CVPR 2024 · similarity 0.90)
- Fine-Grained Post-Training Quantization for Large Vision Language Models with Quantization-Aware Integrated Gradients (CVPR 2026 · 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.
