Large language models demonstrate remarkable capabilities across various domains, especially mathematics and logic reasoning. However, current evaluations overlook physics-based reasoning - a complex task requiring physics theorems and constraints. We present PhysReason, a 1,200-problem benchmark comprising knowledge-based (25%) and reasoning-based (75%) problems, where the latter are divided into three difficulty levels (easy, medium, hard). Notably, problems require an average of 8.1 solution steps, with hard requiring 15.6, reflecting the complexity of physics-based reasoning. We propose the Physics Solution Auto Scoring Framework, incorporating efficient answer-level and comprehensive step-level evaluations. Top-performing models like Deepseek-R1, Gemini-2.0-Flash-Thinking, and o3-mini-high achieve less than 60% on answer-level evaluation, with performance dropping from knowledge questions (75.11%) to hard problems (31.95%). Through step-level evaluation, we identified four key bottlenecks: Physics Theorem Application, Physics Process Understanding, Calculation, and Physics Condition Analysis. These findings position PhysReason as a novel and comprehensive benchmark for evaluating physics-based reasoning capabilities in large language models. Our code and data will be published at https:/dxzxy12138.github.io/PhysReason.

Multimodal Large Language Models (MLLMs) have made significant advancements, demonstrating powerful capabilities in processing and understanding multimodal data. Fine-tuning MLLMs with Federated Learning (FL) allows for expanding the training data scope by including private data sources, thereby enhancing their practical applicability in privacy-sensitive domains. However, current research remains in the early stage, particularly in addressing the \textbf{multimodal heterogeneities} in real-world applications. In this paper, we introduce a benchmark for evaluating various downstream tasks in the federated fine-tuning of MLLMs within multimodal heterogeneous scenarios, laying the groundwork for the research in the field. Our benchmark encompasses two datasets, five comparison baselines, and four multimodal scenarios, incorporating over ten types of modal heterogeneities. To address the challenges posed by modal heterogeneity, we develop a general FedMLLM framework that integrates four representative FL methods alongside two modality-agnostic strategies. Extensive experimental results show that our proposed FL paradigm improves the performance of MLLMs by broadening the range of training data and mitigating multimodal heterogeneity. Code is available at https://github.com/1xbq1/FedMLLM

Complex visual reasoning and question answering (VQA) is a challenging task that requires compositional multi-step processing and higher-level reasoning capabilities beyond the immediate recognition and localization of objects and events. Here, we introduce a fully neural Iterative and Parallel Reasoning Mechanism (IPRM) that combines two distinct forms of computation -- iterative and parallel -- to better address complex VQA scenarios. Specifically, IPRM's "iterative" computation facilitates compositional step-by-step reasoning for scenarios wherein individual operations need to be computed, stored, and recalled dynamically (e.g. when computing the query "determine the color of pen to the left of the child in red t-shirt sitting at the white table"). Meanwhile, its "parallel" computation allows for the simultaneous exploration of different reasoning paths and benefits more robust and efficient execution of operations that are mutually independent (e.g. when counting individual colors for the query: "determine the maximum occurring color amongst all t-shirts"). We design IPRM as a lightweight and fully-differentiable neural module that can be conveniently applied to both transformer and non-transformer vision-language backbones. It notably outperforms prior task-specific methods and transformer-based attention modules across various image and video VQA benchmarks testing distinct complex reasoning capabilities such as compositional spatiotemporal reasoning (AGQA), situational reasoning (STAR), multi-hop reasoning generalization (CLEVR-Humans) and causal event linking (CLEVRER-Humans). Further, IPRM's internal computations can be visualized across reasoning steps, aiding interpretability and diagnosis of its errors.

Most existing studies on few-shot learning focus on unimodal settings, where models are trained to generalize on unseen data using only a small number of labeled examples from the same modality. However, real-world data are inherently multi-modal, and unimodal approaches limit the practical applications of few-shot learning. To address this gap, this paper introduces the Cross-modal Few-Shot Learning (CFSL) task, which aims to recognize instances from multiple modalities when only a few labeled examples are available. This task presents additional challenges compared to classical few-shot learning due to the distinct visual characteristics and structural properties unique to each modality. To tackle these challenges, we propose a Generative Transfer Learning (GTL) framework consisting of two stages: the first stage involves training on abundant unimodal data, and the second stage focuses on transfer learning to adapt to novel data. Our GTL framework jointly estimates the latent shared concept across modalities and in-modality disturbance in both stages, while freezing the generative module during the transfer phase to maintain the stability of the learned representations and prevent overfitting to the limited multi-modal samples. Our finds demonstrate that GTL has superior performance compared to state-of-the-art methods across four distinct multi-modal datasets: Sketchy, TU-Berlin, Mask1K, and SKSF-A. Additionally, the results suggest that the model can estimate latent concepts from vast unimodal data and generalize these concepts to unseen modalities using only a limited number of available samples, much like human cognitive processes.

Causal video question answering (QA) has garnered increasing interest, yet existing datasets often lack depth in causal reasoning. To address this gap, we capitalize on the unique properties of cartoons and construct CausalChaos!, a novel, challenging causal Why-QA dataset built upon the iconic "Tom and Jerry" cartoon series. Cartoons use the principles of animation that allow animators to create expressive, unambiguous causal relationships between events to form a coherent storyline. Utilizing these properties, along with thought-provoking questions and multi-level answers (answer and detailed causal explanation), our questions involve causal chains that interconnect multiple dynamic interactions between characters and visual scenes. These factors demand models to solve more challenging, yet well-defined causal relationships. We also introduce hard incorrect answer mining, including a causally confusing version that is even more challenging. While models perform well, there is much room for improvement, especially, on open-ended answers. We identify more advanced/explicit causal relationship modeling & joint modeling of vision and language as the immediate areas for future efforts to focus upon. Along with the other complementary datasets, our new challenging dataset will pave the way for these developments in the field.

In this work, we introduce the novel concept of visually Connecting Actions and Their Effects (CATE) in video understanding. CATE can have applications in areas like task planning and learning from demonstration. We propose different CATE-based task formulations, such as action selection and action specification, where video understanding models connect actions and effects at semantic and fine-grained levels. We observe that different formulations produce representations capturing intuitive action properties. We also design various baseline models for action selection and action specification. Despite the intuitive nature of the task, we observe that models struggle, and humans outperform them by a large margin. The study aims to establish a foundation for future efforts, showcasing the flexibility and versatility of connecting actions and effects in video understanding, with the hope of inspiring advanced formulations and models.

In this paper, we study multimodal coreference resolution, specifically where a longer descriptive text, i.e., a narration is paired with an image. This poses significant challenges due to fine-grained image-text alignment, inherent ambiguity present in narrative language, and unavailability of large annotated training sets. To tackle these challenges, we present a data efficient semi-supervised approach that utilizes image-narration pairs to resolve coreferences and narrative grounding in a multimodal context. Our approach incorporates losses for both labeled and unlabeled data within a cross-modal framework. Our evaluation shows that the proposed approach outperforms strong baselines both quantitatively and qualitatively, for the tasks of coreference resolution and narrative grounding.

While VideoQA Transformer models demonstrate competitive performance on standard benchmarks, the reasons behind their success are not fully understood. Do these models jointly capture and leverage the rich multimodal structures and dynamics from video and text? Or are they merely exploiting shortcuts to achieve high scores? Hence, we design QUAG (QUadrant AveraGe), a lightweight and non-parametric probe, to critically analyze multimodal representations. QUAG facilitates combined dataset-model study by systematic ablation of model's coupled multimodal understanding during inference. Surprisingly, it demonstrates that the models manage to maintain high performance even under multimodal impairment. We extend QUAG to design "QUAG-attention", a simplistic and lessexpressive replacement of self-attention. We find that the models with QUAGattention achieve similar performance with significantly less mulops without any finetuning. These findings indicate that the current VideoQA benchmarks and metrics do not penalize models that find shortcuts and discount joint multimodal understanding. Motivated by this, we propose the CLAVI (Counterfactual in LAnguage and VIdeo), a diagnostic dataset for coupled multimodal understanding in VideoQA. CLAVI consists of temporal questions and videos that are augmented to curate balanced counterfactuals in language and video domains. We evaluate models on CLAVI and find that all models achieve high performance on multimodal shortcut instances, but most of them have very poor performance on the counterfactual instances that necessitate joint multimodal understanding. Overall, with the multimodal representation analysis using QUAG and diagnostic analysis using CLAVI, we show that many VideoQA models are incapable of learning multimodal representations and that their success on standard datasets is an illusion of joint multimodal understanding. Multimodal learning with videos and language is challenging, despite the shared sequential nature of these modalities, due to their distinct underlying structures. That is, videos exhibit spatio-temporal dynamics in the pixel space, whereas language representation is composed of the syntax and semantics of word sequences. Hence, tasks like Video Question Answering (VideoQA) (Zhong et al., 2022) are difficult as they necessitate the model to acquire accurate representations of both the modalities and establish meaningful connections between them. Transformers have demonstrated exceptional performance on VideoQA benchmarks (Zhong et al., 2022).

The problem of anticipating human actions is an inherently uncertain one. However, we can reduce this uncertainty if we have a sense of the goal that the actor is trying to achieve. Here, we present an action anticipation model that leverages goal information for the purpose of reducing the uncertainty in future predictions. Since we do not possess goal information or the observed actions during inference, we resort to visual representation to encapsulate information about both actions and goals. Through this, we derive a novel concept called abstract goal which is conditioned on observed sequences of visual features for action anticipation. We design the abstract goal as a distribution whose parameters are estimated using a variational recurrent network. We sample multiple candidates for the next action and introduce a goal consistency measure to determine the best candidate that follows from the abstract goal. Our method obtains impressive results on the very challenging Epic-Kitchens55 (EK55), EK100, and EGTEA Gaze+ datasets. We obtain absolute improvements of +13.69, +11.24, and +5.19 for Top-1 verb, Top-1 noun, and Top-1 action anticipation accuracy respectively over prior state-of-the-art methods for seen kitchens (S1) of EK55. Similarly, we also obtain significant improvements in the unseen kitchens (S2) set for Top-1 verb (+10.75), noun (+5.84) and action (+2.87) anticipation. Similar trend is observed for EGTEA Gaze+ dataset, where absolute improvement of +9.9, +13.1 and +6.8 is obtained for noun, verb, and action anticipation. It is through the submission of this paper that our method is currently the new state-of-the-art for action anticipation in EK55 and EGTEA Gaze+ https://competitions.codalab.org/competitions/20071#results Code available at https://github.com/debadityaroy/Abstract_Goal

Coreference resolution aims to identify words and phrases which refer to same entity in a text, a core task in natural language processing. In this paper, we extend this task to resolving coreferences in long-form narrations of visual scenes. First we introduce a new dataset with annotated coreference chains and their bounding boxes, as most existing image-text datasets only contain short sentences without coreferring expressions or labeled chains. We propose a new technique that learns to identify coreference chains using weak supervision, only from image-text pairs and a regularization using prior linguistic knowledge. Our model yields large performance gains over several strong baselines in resolving coreferences. We also show that coreference resolution helps improving grounding narratives in images.