8 papers
Metacognition as Reward: Reinforcing LLM Reasoning via Knowledge and Regulation Signals
Sirui Chen, Lei Xu, Yuying Zhao +6
Recent RL methods have substantially improved the reasoning abilities of LLMs. Existing reward designs mainly follow two paradigms: (1) Reinforcement learning with verifiable rewar…
Generalized Logit Adjustment: Calibrating Fine-tuned Models by Removing Label Bias in Foundation Models
Beier Zhu, Kaihua Tang, Qianru Sun +1
Foundation models like CLIP allow zero-shot transfer on various tasks without additional training data. Yet, the zero-shot performance is less competitive than a fully supervised o…
Debiased Fine-Tuning for Vision-language Models by Prompt Regularization
Beier Zhu, Yulei Niu, Saeil Lee +2
We present a new paradigm for fine-tuning large-scale visionlanguage pre-trained models on downstream task, dubbed Prompt Regularization (ProReg). Different from traditional fine-t…
Prompt-aligned Gradient for Prompt Tuning
Beier Zhu, Yulei Niu, Yucheng Han +2
Thanks to the large pre-trained vision-language models (VLMs) like CLIP, we can craft a zero-shot classifier by "prompt", e.g., the confidence score of an image being "[CLASS]" can…
Unsupervised Visual Chain-of-Thought Reasoning via Preference Optimization
Kesen Zhao, Beier Zhu, Qianru Sun +1
Chain-of-thought (CoT) reasoning greatly improves the interpretability and problem-solving abilities of multimodal large language models (MLLMs). However, existing approaches are f…
Robust Fine-tuning of Zero-shot Models via Variance Reduction
Beier Zhu, Jiequan Cui, Hanwang Zhang
When fine-tuning zero-shot models like CLIP, our desideratum is for the fine-tuned model to excel in both in-distribution (ID) and out-of-distribution (OOD). Recently, ensemble-bas…