Vision-Language Models (VLMs) have demonstrated remarkable capabilities in interpreting visual layouts and text. However, a significant challenge remains in their ability to interpret robustly and reason over multi-tabular data presented as images, a common occurrence in real-world scenarios like web pages and digital documents. Existing benchmarks typically address single tables or non-visual data (text/structured). This leaves a critical gap: they don’t assess the ability to parse diverse table images, correlate information across them, and perform multi-hop reasoning on the combined visual data. To bridge this evaluation gap, we introduce MTabVQA, a novel benchmark specifically designed for multi-tabular visual question answering. MTabVQA comprises 3,745 complex question-answer pairs that necessitate multi-hop reasoning across several visually rendered table images. We provide extensive benchmark results for state-of-the-art VLMs on MTabVQA, revealing significant performance limitations. We further investigate post-training techniques to enhance these reasoning abilities and release MTabVQA-Instruct, a large-scale instruction-tuning dataset. Our experiments show that fine-tuning VLMs with MTabVQA-Instruct substantially improves their performance on visual multi-tabular reasoning. Code and dataset are available online: .

This paper presents our submission to the Fin-DBQA shared task at the 9th FinNLP workshop. The task involves answering finance-focused questions in a multi-turn environment, requiring step-by-step reasoning and Python code generation. We propose a novel approach to tackle this multidimensional problem by pre-processing the data into tidy data format so that each column represents a variable and each row an observation. Our experiments demonstrate that using the tidy data format allows all models to surpass SOTA, with GPT-4o achieving a 50.62% accuracy on the DBQR-QA benchmark achieving second place on the shared task leaderboard. These findings suggest that transforming data into the tidy data format enhances reasoning capabilities, reduces syntax errors, and improves performance on table-reasoning QA tasks. The code is available online.

Scientific charts often encapsulate the core findings of research papers, making the ability to answer questions about these charts highly valuable. This paper explores recent advancements in scientific chart visual question answering (VQA) enabled by large Vision Language Models (VLMs) and newly curated datasets. As part of the SciVQA shared task from the 5th Workshop on Scholarly Document Processing, we develop and evaluate multimodal Systems capable of answering diverse question types - including multiple-choice, yes/no, unanswerable, and infinite answer set questions - based on chart images extracted from scientific literature. We investigate the effects of zero-shot and one-shot prompting, as well as supervised fine-tuning (SFT), on the performance of Qwen2.5-VL models (7B and 32B variants). We also tried to include more training data from domain-specific datasets (SpiQA and ArXivQA). Our fine-tuned Qwen2.5-VL 32B model achieves a substantial improvement over the GPT-4o-mini baseline and reaches the 4th place in the shared task, highlighting the effectiveness of domain-specific fine-tuning. We published the code for the experiments.

Large Language Models (LLMs) such as ChatGPT, GitHub Copilot, Llama, or Mistral assist programmers as copilots and knowledge sources to make the coding process faster and more efficient. This paper aims to improve the copilot performance by implementing different self-alignment processes and retrieval-augmented generation (RAG) pipelines, as well as their combination. To test the effectiveness of all approaches, we create a dataset and apply a model-based evaluation, using LLM as a judge. It is designed to check the model’s abilities to understand the source code semantics, the dependency between files, and the overall meta-information about the repository. We also compare our approach with other existing solutions, e.g. ChatGPT-3.5, and evaluate on the existing benchmarks. Code and dataset are available online (https://anonymous.4open.science/r/ma_llm-382D).