paper-with-me

Papers

Prefix Conditioning Unifies Language and Label Supervision

2022-06-02 · CVPR 2023 1 · Kuniaki Saito, Kihyuk Sohn, Xiang Zhang, Chun-Liang Li, Chen-Yu Lee, Kate Saenko, Tomas Pfister

Image-classification datasets have been used to pretrain image recognition models. Recently, web-scale image-caption datasets have emerged as a source of powerful pretraining alternative. Image-caption datasets are more ``open-domain'', containing a wider variety of scene types and vocabulary words than traditional classification datasets, and models trained on these datasets have demonstrated strong performance on few- and zero-shot recognition tasks. When naively unifying image-classification and -caption dataset, we show that such dataset biases negatively affect pre-training by reducing the generalizability of learned representations and thus jeopardizing zero-shot performance since the unification can tailor the model for the classification dataset, making it vulnerable to the distribution shift from the dataset. In this work, we address the problem by disentangling the dataset bias using prefix tokens that inform a language encoder of the type of the input dataset (e.g., image-classification or caption) at training time. This approach allows the language encoder to share the knowledge from two datasets as well as switch the mode of feature extraction, i.e., image-classification dataset or image-caption dataset tailored mode, where we use image-caption mode in the zero-shot evaluation. Our method is generic and can be easily integrated into existing VL pre-training objectives such as CLIP or UniCL. In experiments, we show that this simple technique improves the performance in zero-shot image recognition accuracy and robustness to the image-level distribution shift.

📄 PDF Abstract BibTeX arXiv:2206.01125

Code (0)

등록된 구현이 없습니다.

Tasks

ClassificationContrastive Learningimage-classificationImage ClassificationZero-Shot Learning

Methods 이 논문이 사용한 방법론

CLIP Contrastive Language-Image Pre-training (CLIP), consisting of a simplified version of ConVIRT trained from scratch, is an efficient method of image representation learning…
Contrastive Learning 설명 없음

Similar Papers 제목 키워드 기반

Learning from the Self-future: On-policy Self-distillation for dLLMs

2026-06-16 · Yifu Luo, Zeyu Chen, Haoyu Wang, Xinhao Hu 외 arxiv

On-policy self-distillation (OPSD) has proven effective for post-training large language models (LLMs), yet its application to diffusion LLMs (dLLMs) remains unexplored. Existing OPSD methods are inherently autoregressiv…

From Correctness to Utility: Gain-Based Prefix Evaluation for LLM Reasoning

2026-06-05 · Yuhang Zhou, Yixin Cao, Guangnan Ye arxiv

Reasoning prefixes shape the future trajectory of LLM problem solving, yet existing process reward models usually evaluate them through local step correctness. We argue that correctness is a useful but indirect proxy for…

Reinforcement LearningMathematical Reasoning

Training Reasoning Models on Saturated Problems via Failure-Prefix Conditioning

2026-01-28 · Minwu Kim, Safal Shrestha, Anubhav Shrestha, Keith Ross arxiv

As Reinforcement Learning with Verifiable Rewards (RLVR) substantially improves the reasoning abilities of large language models (LLMs), a new bottleneck emerges: more training problems become saturated, that is, the LLM…

Reinforcement Learning

ROSD: Reflective On-Policy Self-Distillation for Language Model Reasoning across Domains

2026-05-27 · Ziqi Zhao, Xinyu Ma, Liu Yang, Yujie Feng 외 arxiv

On-policy self-distillation (OPSD) improves the reasoning performance of large language models (LLMs) by providing dense token-level supervision for on-policy rollouts. However, existing OPSD methods often yield limited …

Domain Generalization

Spec-AUF: Accept-Until-Fail Training under Train-Inference Misalignment for Masked Block Drafters

2026-07-02 · Tianjian Yang, Meng Li arxiv

Speculative decoding accelerates autoregressive generation by drafting a block of tokens that the target model verifies left-to-right, committing only the longest accepted prefix. Block (DLM-style) drafters predict the w…