Large Language Models (LLMs) have shown remarkable success in multi-hop question-answering (M-QA) due to their advanced reasoning capabilities. However, the influence of reasoning structures on their performance remains underexplored, primarily due to the lack of M-QA datasets that explicitly encode the reasoning pathways underlying each question-answer pair. To address this gap, we introduce the reasoning graph-structured question answering dataset (GRS-QA), which provides both semantic contexts and reasoning structures for the QA pairs. Unlike existing M-QA datasets, GRS-QA explicitly captures intricate reasoning pathways through reasoning graphs, where nodes correspond to textual contexts and edges denote logical flows. Using GRS-QA, we systematically evaluate LLM performance across varying context structures, prompting styles, and data domains. Our empirical analysis reveals that LLMs perform differently based on the reasoning structure, context, and prompting styles, indicating their varying ability to leverage graph-structured knowledge. Notably, providing explicit reasoning guidance proves more effective than supplying contextual information alone.

Recent web agents use world models for test-time action selection by sampling candidate actions, predicting the resulting web states, and ranking them with a ranker model or a Process Reward Model (PRM). These world models are typically trained via supervised next-state prediction to generate fixed representations like HTML or AXTree snapshots. However, this objective is misaligned with the downstream ranker, which relies on predicted states being discriminative across candidates to accurately score them. To address this, we introduce predicted-state matching, a training objective where the predicted representation must distinguish the true resulting state from those reached by alternative actions. We train these models using a branching web-agent dataset derived from WebArena Go-Browse trajectories, where every decision point contains multiple alternative actions and their resulting states. Experiments on our held-out predicted-state matching benchmark show that our approach outperforms world models trained with supervised next-state prediction. We further show that our approach improves PRM-style action ranking on WebPRMBench compared with action-only PRMs and PRMs augmented with supervised-next-state world models. Finally, on WebArena-Lite, using our world model for test-time action selection improves end-to-end task success. Our project page is available at: https://dhruvpendharkar.github.io/dwm/.

This paper presents our submission to the SemEval 2024 Task 5: The Legal Argument Reasoning Task in Civil Procedure. We present two approaches to solving the task of legal answer validation, given an introduction to the case, a question and an answer candidate. Firstly, we fine-tuned pre-trained BERT-based models and found that models trained on domain knowledge perform better. Secondly, we performed few-shot prompting on GPT models and found that reformulating the answer validation task to be a multiple-choice QA task remarkably improves the performance of the model. Our best submission is a BERT-based model that achieved the 7th place out of 20.