arXiv subjects

Akshit Sinha

Publications and source records attributed to Akshit Sinha.

5 recordsLinked to original sources

arXiv2026

The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs

Does continued scaling of large language models (LLMs) yield diminishing returns? In this work, we show that short-task benchmarks may give an illusion of slowing progress, as even marginal gains in single-step accuracy can compound into exponential improvements in the length of tasks a model can successfully complete. Then, we argue that failures of LLMs when simple tasks are made longer arise from mistakes in execution, rather than an inability to reason. So, we propose isolating execution capability, by explicitly providing the knowledge and plan needed to solve a long-horizon task. First, we find that larger models can correctly execute significantly more turns even when small models have near-perfect single-turn accuracy. We then observe that the per-step accuracy of models degrades as the number of steps increases. This is not just due to long-context limitations -- curiously, we observe a self-conditioning effect -- models become more likely to make mistakes when the context contains their errors from prior turns. Self-conditioning does not reduce by just scaling the model size. But, we find that thinking mitigates self-conditioning, and also enables execution of much longer tasks in a single turn. We conclude by benchmarking frontier thinking models on the length of tasks they can execute in a single turn. Overall, by focusing on the ability to execute, we hope to reconcile debates on how LLMs can solve complex reasoning problems yet fail at simple tasks when made longer, and highlight the massive benefits of scaling model size and sequential test-time compute for long-horizon tasks.

cs.AI↗

arXiv2025

A Graph Talks, But Who's Listening? Rethinking Evaluations for Graph-Language Models

Developments in Graph-Language Models (GLMs) aim to integrate the structural reasoning capabilities of Graph Neural Networks (GNNs) with the semantic understanding of Large Language Models (LLMs). However, we demonstrate that current evaluation benchmarks for GLMs, which are primarily repurposed node-level classification datasets, are insufficient to assess multimodal reasoning. Our analysis reveals that strong performance on these benchmarks is achievable using unimodal information alone, suggesting that they do not necessitate graph-language integration. To address this evaluation gap, we introduce the CLEGR(Compositional Language-Graph Reasoning) benchmark, designed to evaluate multimodal reasoning at various complexity levels. Our benchmark employs a synthetic graph generation pipeline paired with questions that require joint reasoning over structure and textual semantics. We perform a thorough evaluation of representative GLM architectures and find that soft-prompted LLM baselines perform on par with GLMs that incorporate a full GNN backbone. This result calls into question the architectural necessity of incorporating graph structure into LLMs. We further show that GLMs exhibit significant performance degradation in tasks that require structural reasoning. These findings highlight limitations in the graph reasoning capabilities of current GLMs and provide a foundation for advancing the community toward explicit multimodal reasoning involving graph structure and language.

cs.CL↗

arXiv2025

A Cognac Shot To Forget Bad Memories: Corrective Unlearning for Graph Neural Networks

Graph Neural Networks (GNNs) are increasingly being used for a variety of ML applications on graph data. Because graph data does not follow the independently and identically distributed (i.i.d.) assumption, adversarial manipulations or incorrect data can propagate to other data points through message passing, which deteriorates the model's performance. To allow model developers to remove the adverse effects of manipulated entities from a trained GNN, we study the recently formulated problem of Corrective Unlearning. We find that current graph unlearning methods fail to unlearn the effect of manipulations even when the whole manipulated set is known. We introduce a new graph unlearning method, Cognac, which can unlearn the effect of the manipulation set even when only 5% of it is identified. It recovers most of the performance of a strong oracle with fully corrected training data, even beating retraining from scratch without the deletion set while being 8x more efficient. We hope our work assists GNN developers in mitigating harmful effects caused by issues in real-world data, post-training. Our code is publicly available at https://github.com/cognac-gnn-unlearning/corrective-unlearning-for-gnns

cs.LG↗

arXiv2025

Topo Goes Political: TDA-Based Controversy Detection in Imbalanced Reddit Political Data

The detection of controversial content in political discussions on the Internet is a critical challenge in maintaining healthy digital discourse. Unlike much of the existing literature that relies on synthetically balanced data, our work preserves the natural distribution of controversial and non-controversial posts. This real-world imbalance highlights a core challenge that needs to be addressed for practical deployment. Our study re-evaluates well-established methods for detecting controversial content. We curate our own dataset focusing on the Indian political context that preserves the natural distribution of controversial content, with only 12.9% of the posts in our dataset being controversial. This disparity reflects the true imbalance in real-world political discussions and highlights a critical limitation in the existing evaluation methods. Benchmarking on datasets that model data imbalance is vital for ensuring real-world applicability. Thus, in this work, (i) we release our dataset, with an emphasis on class imbalance, that focuses on the Indian political context, (ii) we evaluate existing methods from this domain on this dataset and demonstrate their limitations in the imbalanced setting, (iii) we introduce an intuitive metric to measure a model's robustness to class imbalance, (iv) we also incorporate ideas from the domain of Topological Data Analysis, specifically Persistent Homology, to curate features that provide richer representations of the data. Furthermore, we benchmark models trained with topological features against established baselines.

cs.SI↗

arXiv2025

Higher Order Structures For Graph Explanations

Graph Neural Networks (GNNs) have emerged as powerful tools for learning representations of graph-structured data, demonstrating remarkable performance across various tasks. Recognising their importance, there has been extensive research focused on explaining GNN predictions, aiming to enhance their interpretability and trustworthiness. However, GNNs and their explainers face a notable challenge: graphs are primarily designed to model pair-wise relationships between nodes, which can make it tough to capture higher-order, multi-node interactions. This characteristic can pose difficulties for existing explainers in fully representing multi-node relationships. To address this gap, we present Framework For Higher-Order Representations In Graph Explanations (FORGE), a framework that enables graph explainers to capture such interactions by incorporating higher-order structures, resulting in more accurate and faithful explanations. Extensive evaluation shows that on average real-world datasets from the GraphXAI benchmark and synthetic datasets across various graph explainers, FORGE improves average explanation accuracy by 1.9x and 2.25x, respectively. We perform ablation studies to confirm the importance of higher-order relations in improving explanations, while our scalability analysis demonstrates FORGE's efficacy on large graphs.

cs.LG↗