Profiles2020-2026
Pengfei Cao
University of Chinese Academy of Sciences
49Papers
0Last author
39Unique co-authors
6First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers49
- MMR-V: What's Left Unsaid? A Benchmark for Multimodal Deep Reasoning in VideosAuthor 6 of 9ICLR 2026 · 22 citations
- Omni-Reward: Towards Generalist Omni-Modal Reward Modeling with Free-Form PreferencesAuthor 5 of 8ICLR 2026 · 11 citations
- MMR-Life: Piecing Together Real-life Scenes for Multimodal Multi-image ReasoningAuthor 5 of 8ICLR 2026 · 7 citations
- Task-Stratified Knowledge Scaling Laws for Post-Training Quantized Large Language ModelsAuthor 2 of 7ACL Findings 2026 · 4 citations
- Fixing the Broken Compass: Diagnosing and Improving Inference-Time Reward ModelingAuthor 2 of 9ICLR 2026 · 3 citations
- Bias-Restrained Prefix Representation Finetuning for Mathematical ReasoningAuthor 2 of 6AAAI 2026
- Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task PlanningAuthor 3 of 6ACL 2026
- From Signal Degradation to Computation Collapse: Uncovering the Two Failure Modes of LLM QuantizationAuthor 2 of 7ACL Findings 2026
- Learning How to Remember: A Meta-Cognitive Management Method for Structured and Transferable Agent MemoryAuthor 2 of 7ACL Findings 2026
- Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot DoAuthor 5 of 8ACL 2026
- Towards Atoms of Large Language ModelsAuthor 2 of 5ICML 2026
- Towards Robust Knowledge Unlearning: An Adversarial Framework for Assessing and Improving Unlearning Robustness in Large Language ModelsAuthor 3 of 6AAAI 2025 · 26 citations
- Agent-RewardBench: Towards a Unified Benchmark for Reward Modeling across Perception, Planning, and Safety in Real-World Multimodal AgentsAuthor 3 of 6ACL 2025 · 13 citations
- Knowledge in Superposition: Unveiling the Failures of Lifelong Knowledge Editing for Large Language ModelsAuthor 2 of 5AAAI 2025 · 9 citations
- CITI: Enhancing Tool Utilizing Ability in Large Language Models Without Sacrificing General PerformanceAuthor 2 of 7AAAI 2025 · 6 citations
- Towards Better Chain-of-Thought: A Reflection on Effectiveness and FaithfulnessAuthor 2 of 8ACL Findings 2025 · 5 citations
- Revealing the Deceptiveness of Knowledge Editing: A Mechanistic Analysis of Superficial EditingAuthor 2 of 5ACL 2025 · 2 citations
- RAG-RewardBench: Benchmarking Reward Models in Retrieval Augmented Generation for Preference AlignmentAuthor 4 of 10ACL Findings 2025 · 1 citation
- A Troublemaker with Contagious Jailbreak Makes Chaos in Honest TownsAuthor 2 of 6ACL 2025
- Cracking Factual Knowledge: A Comprehensive Analysis of Degenerate Knowledge Neurons in Large Language ModelsAuthor 2 of 7ACL 2025
- Evaluating Personalized Tool-Augmented LLMs from the Perspectives of Personalization and ProactivityAuthor 2 of 7ACL 2025
- Knowledge Localization: Mission Not Accomplished? Enter Query Localization!Author 2 of 5ICLR 2025
- M2Edit: Locate and Edit Multi-Granularity Knowledge in Multimodal Large Language ModelAuthor 2 of 8EMNLP 2025
- MIRAGE: Evaluating and Explaining Inductive Reasoning Process in Language ModelsAuthor 2 of 6ICLR 2025
- The Knowledge Microscope: Features as Better Analytical Lenses than NeuronsAuthor 2 of 4ACL 2025
- Journey to the Center of the Knowledge Neurons: Discoveries of Language-Independent Knowledge Neurons and Degenerate Knowledge NeuronsAuthor 2 of 5AAAI 2024 · 64 citations
- AgentsCourt: Building Judicial Decision-Making Agents with Court Debate Simulation and Legal Knowledge AugmentationAuthor 2 of 9EMNLP Findings 2024 · 20 citations
- Unlocking the Future: Exploring Look-Ahead Planning Mechanistic Interpretability in Large Language ModelsAuthor 2 of 6EMNLP 2024 · 20 citations
- LINKED: Eliciting, Filtering and Integrating Knowledge in Large Language Model for Commonsense ReasoningAuthor 2 of 9EMNLP Findings 2024 · 12 citations
- WilKE: Wise-Layer Knowledge Editor for Lifelong Knowledge EditingAuthor 2 of 5ACL Findings 2024 · 10 citations
Showing the 30 most recent indexed papers.
