A Closer Look at the Few-Shot Adaptation of Large Vision-Language Models

Keyphrases

• Adaptation 33%
• Augmented Lagrangian Method 33%
• Distributional Drift 33%
• Downstream Task 33%
• False Assumption 33%
• Few-shot Adaptation 100%
• Grid Search 33%
• Hyperparameters 66%
• Labeled Sample 33%
• Large Corpora 33%
• Large Vision-Language Models 100%
• Limited Labeled Samples 33%
• Linear Probing 100%
• Optimal Hyperparameters 33%
• Parameter Transfer Learning 66%
• Transfer Learning Models 33%
• Validation Case 33%
• Validation Set 33%
• Vision-Language Models 33%
• Zero-shot Prediction 33%

Computer Science

• Language Modeling 100%
• Leaning Parameter 100%
• Learning Approach 50%
• Transfer Learning 100%
• Validation Set 50%
• Zero-Shot Learning 50%