Author name cluster

Pingyue Zhang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

3 papers

1 author row

AAAI Conference 2026 Conference Paper

WorldAgen: Unified State-Action Prediction with Test-Time World Model Training

Chi Wan
Kangrui Wang
Yuan Si
Pingyue Zhang
Manling Li

How can vision-language-action (VLA) models adapt to new environments where world dynamics shift? While recent research has combined world modeling and action prediction to improve VLA performance, existing methods largely rely on pretraining in static datasets, without mechanisms for active adaptation to new environments. As a result, these models often fail to generalize when deployed in unseen scenarios with novel object configurations or dynamics. We present WorldAgen, a unified framework that jointly learns world modeling and action prediction while enabling test-time training (TTT) to adapt to new environments. WorldAgen employs a shared Transformer backbone with two heads: (1) a world-model head that predicts future states from past state-action trajectories, and (2) an agent-model head that predicts actions conditioned on task instructions. During test time, WorldAgen samples exploratory actions, collects ground-truth state transitions, and performs lightweight TTT updates to refine its world model. This adaptation improves the model's understanding to the environments and leads to more accurate action predictions. Experiments on the CALVIN and LIBERO benchmarks demonstrate that our baseline model achieves comparable, and in some cases superior, performance to current state-of-the-art approaches. Moreover, with TTT on a small number of samples, our method surpasses existing state-of-the-art models, highlighting the effectiveness of adapting world models at inference time.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

VAGEN: Reinforcing World Model Reasoning for Multi-Turn VLM Agents

Kangrui Wang
Pingyue Zhang
Zihan Wang
Yaning Gao
Linjie Li
Qineng Wang
Hanyang Chen
Yiping Lu

A major challenge in training VLM agents, compared to LLM agents, is that states shift from simple texts to complex visual observations, which introduces partial observability and demands robust world modeling. We ask: can VLM agents build internal world models through explicit visual state reasoning? In this work, we architecturally enforce and reward VLM agent’s reasoning process via reinforcement learning (RL), formulating the problem as a Partially Observable Markov Decision Process (POMDP). We demonstrate that structuring agent’s reasoning into StateEstimation (“what is the current state? ”) and TransitionModeling (“what is next? ”) is critical by studying five reasoning strategies. Investigating how agents should ground visual states and represent these internal beliefs, we reveal the optimal representations are task-dependent: Natural Language excels at capturing semantic relationships for general tasks, while Structured formats are essential for high-precision manipulation. These insights motivate our approach to reward shaping and credit assignment. We leverage a WorldModeling Reward to densely rewards the agent’s turn-by-turn state predictions, while our Bi-Level General Advantage Estimation (Bi-Level GAE) enables turn-aware credit assignment. Through such world model reasoning, we enable a 3B model to achieve performance of 0. 82 on a set of five diverse agent tasks, nearly 3× improvement over its untrained counterpart (0. 21) and surpassing proprietary reasoning models like GPT-5 (0. 75), Gemini 2. 5 Pro (0. 67) and Claude 4. 5 (0. 62). All experiments are supported by our VAGEN framework, a scalable system for training and analyzing multi-turn VLM agents across diverse visual environments

PDF Details

AAAI Conference 2024 Conference Paper

Multi-Label Supervised Contrastive Learning

Pingyue Zhang
Mengyue Wu

Multi-label classification is an arduous problem given the complication in label correlation. Whilst sharing a common goal with contrastive learning in utilizing correlations for representation learning, how to better leverage label information remains challenging. Previous endeavors include extracting label-level presentations or mapping labels to an embedding space, overlooking the correlation between multiple labels. It exhibits a great ambiguity in determining positive samples with different extent of label overlap between samples and integrating such relations in loss functions. In our work, we propose Multi-Label Supervised Contrastive learning (MulSupCon) with a novel contrastive loss function to adjust weights based on how much overlap one sample shares with the anchor. By analyzing gradients, we explain why our method performs better under multi-label circumstances. To evaluate, we conduct direct classification and transfer learning on several multi-label datasets, including widely-used image datasets such as MS-COCO and NUS-WIDE. Validation indicates that our method outperforms the traditional multi-label classification method and shows a competitive performance when comparing to other existing approaches.

PDF Details DOI