1 citations · 2 across the 22 of their papers we have counts for
22 papers
GeoMathCode: Understanding Interleaved Math-Code Reasoning for Geometry Problem Solving
Yingji Zhang, Yong Dai, André Freitas
Mathematical reasoning is a hallmark of human intelligence, requiring logical deduction, symbolic manipulation, and abstract thinking. Recent multimodal large language models (MLLM…
Pelican-Unify 1.0: A Unified Embodied Intelligence Model for Understanding, Reasoning, Imagination and Action
Yi Zhang, Yinda Chen, Che Liu +26
We present Pelican-Unify 1.0, the first embodied foundation model trained according to the principle of unification. Pelican-Unify 1.0 uses a single VLM as a unified understanding…
EPIC-Bench: A Perception-Centric Benchmark for Fine-Grained Embodied Visual Grounding in Vision-Language Models
Haozhe Shan, Xiancong Ren, Han Dong +10
While large vision-language models (VLMs) are increasingly adopted as the perceptual backbone for embodied agents, their fundamental vision-language grounding (VLG) capability in e…
VLA-Trace: Diagnosing Vision-Language-Action Models through Representation and Behavior Tracing
Haoyuan Shi, Xiancong Ren, Yingji Zhang +9
Understanding how Vision-Language-Action (VLA) models transform multimodal knowledge into embodied control remains an open challenge. We present VLA-Trace, a progressive diagnostic…
A2Eval: Agentic and Automated Evaluation for Embodied Brain
Shuai Zhang, Jiayu Hu, Zijie Chen +9
Current embodied VLM evaluation relies on static, expert-defined, manually annotated benchmarks that exhibit severe redundancy and coverage imbalance. This labor intensive paradigm…
Formal Semantic Control over Language Models
Yingji Zhang
This thesis advances semantic representation learning to render language representations or models more semantically and geometrically interpretable, and to enable localised, quasi…
Bridging VLMs and Embodied Intelligence with Deliberate Practice Policy Optimization
Yi Zhang, Che Liu, Xiancong Ren +17
Developing a universal and versatile embodied intelligence system presents two primary challenges: the critical embodied data bottleneck, where real-world data is scarce and expens…
Wanderland: Geometrically Grounded Simulation for Open-World Embodied AI
Xinhao Liu, Jiaqi Li, Youming Deng +7
Reproducible closed-loop evaluation remains a major bottleneck in Embodied AI such as visual navigation. A promising path forward is high-fidelity simulation that combines photorea…
Beyond Surface Alignment: Rebuilding LLMs Safety Mechanism via Probabilistically Ablating Refusal Direction
Yuanbo Xie, Yingjie Zhang, Tianyun Liu +2
Jailbreak attacks pose persistent threats to large language models (LLMs). Current safety alignment methods have attempted to address these issues, but they experience two signific…
Web-CogReasoner: Towards Multimodal Knowledge-Induced Cognitive Reasoning for Web Agents
Yuhan Guo, Cong Guo, Aiwen Sun +12
Multimodal large-scale models have significantly advanced the development of web agents, enabling perception and interaction with digital environments akin to human cognition. In t…
Bridging Compositional and Distributional Semantics: A Survey on Latent Semantic Geometry via AutoEncoder
Yingji Zhang, Danilo S. Carvalho, André Freitas
Integrating compositional and symbolic properties into current distributional semantic spaces can enhance the interpretability, controllability, compositionality, and generalisatio…
Learning to Disentangle Latent Reasoning Rules with Language VAEs: A Systematic Study
Yingji Zhang, Marco Valentino, Danilo S. Carvalho +1
Incorporating explicit reasoning rules within the latent space of language models (LMs) offers a promising pathway to enhance generalisation, interpretability, and controllability.…
LangVAE and LangSpace: Building and Probing for Language Model VAEs
Danilo S. Carvalho, Yingji Zhang, Harriet Unsworth +1
We present LangVAE, a novel framework for modular construction of variational autoencoders (VAEs) on top of pre-trained large language models (LLMs). Such language model VAEs can e…
Nexus: An Omni-Perceptive And -Interactive Model for Language, Audio, And Vision
Che Liu, Yingji Zhang, Dong Zhang +13
This work proposes an industry-level omni-modal large language model (LLM) pipeline that integrates auditory, visual, and linguistic modalities to overcome challenges such as limit…
Does Table Source Matter? Benchmarking and Improving Multimodal Scientific Table Understanding and Reasoning
Bohao Yang, Yingji Zhang, Dong Liu +2
Recent large language models (LLMs) have advanced table understanding capabilities but rely on converting tables into text sequences. While multimodal large language models (MLLMs)…
Improving Semantic Control in Discrete Latent Spaces with Transformer Quantized Variational Autoencoders
Yingji Zhang, Danilo S. Carvalho, Marco Valentino +2
Achieving precise semantic control over the latent spaces of Variational AutoEncoders (VAEs) holds significant value for downstream tasks in NLP as the underlying generative mechan…
LlaMaVAE: Guiding Large Language Model Generation via Continuous Latent Sentence Spaces
Yingji Zhang, Danilo S. Carvalho, Ian Pratt-Hartmann +1
Deep generative neural networks, such as Variational AutoEncoders (VAEs), offer an opportunity to better understand and control language models from the perspective of sentence-lev…
Graph-Induced Syntactic-Semantic Spaces in Transformer-Based Variational AutoEncoders
Yingji Zhang, Marco Valentino, Danilo S. Carvalho +2
The injection of syntactic information in Variational AutoEncoders (VAEs) has been shown to result in an overall improvement of performances and generalisation. An effective strate…
Towards Controllable Natural Language Inference through Lexical Inference Types
Yingji Zhang, Danilo S. Carvalho, Ian Pratt-Hartmann +1
Explainable natural language inference aims to provide a mechanism to produce explanatory (abductive) inference chains which ground claims to their supporting premises. A recent co…
Learning Disentangled Semantic Spaces of Explanations via Invertible Neural Networks
Yingji Zhang, Danilo S. Carvalho, André Freitas
Disentangled latent spaces usually have better semantic separability and geometrical properties, which leads to better interpretability and more controllable data generation. While…
Quasi-symbolic Semantic Geometry over Transformer-based Variational AutoEncoder
Yingji Zhang, Danilo S. Carvalho, André Freitas
Formal/symbolic semantics can provide canonical, rigid controllability and interpretability to sentence representations due to their \textit{localisation} or \textit{composition} p…
Learning Disentangled Representations for Natural Language Definitions
Danilo S. Carvalho, Giangiacomo Mercatali, Yingji Zhang +1
Disentangling the encodings of neural models is a fundamental aspect for improving interpretability, semantic control and downstream task performance in Natural Language Processing…