ServiceNow AI Research

1

AlignVLM: Bridging Vision and Language Latent Spaces for Multimodal Document Understanding
AlignVLM: Bridging Vision and Language Latent Spaces for Multimodal Document Understanding
Aligning visual features with language embeddings is a key challenge in vision-language models (VLMs). The performance of such models …
Neural Information Processing Systems (NeurIPS),  2025.
Paper Cite Code Video
Causal Differentiating Concepts: Interpreting LM Behavior via Causal Representation Learning
Causal Differentiating Concepts: Interpreting LM Behavior via Causal Representation Learning
Language model activations entangle concepts that mediate their behavior, making it difficult to interpret these factors, which has …
Neural Information Processing Systems (NeurIPS),  2025.
Paper Cite
How to Train Your LLM Web Agent: A Statistical Diagnosis
How to Train Your LLM Web Agent: A Statistical Diagnosis

Large language model (LLM) agents for web interfaces have advanced rapidly, yet open-source systems still lag behind proprietary …

Neural Information Processing Systems (NeurIPS),  2025.
Paper Cite
Rendering-Aware Reinforcement Learning for Vector Graphics Generation
Rendering-Aware Reinforcement Learning for Vector Graphics Generation
Scalable Vector Graphics (SVG) offer a powerful format for representing visual designs as interpretable code. Recent advances in …
Neural Information Processing Systems (NeurIPS),  2025.
Paper Cite
The Promise of RL for Autoregressive Image Editing
The Promise of RL for Autoregressive Image Editing
While image generation techniques are now capable of producing high quality images that respect prompts which span multiple sentences, …
Neural Information Processing Systems (NeurIPS),  2025.
Paper Cite Code
ColMate: Contrastive Late Interaction and Masked Text for Multimodal Document Retrieval
ColMate: Contrastive Late Interaction and Masked Text for Multimodal Document Retrieval
Retrieval-augmented generation has proven practical when models require specialized knowledge or access to the latest data. However, …
Conference on Empirical Methods in Natural Language Processing (EMNLP),  2025.
Paper Cite
FM2DS: Few-Shot Multimodal Multihop Data Synthesis with Knowledge Distillation for Question Answering
FM2DS: Few-Shot Multimodal Multihop Data Synthesis with Knowledge Distillation for Question Answering
We introduce DRBench, a benchmark for evaluating AI agents on complex, open-ended enterprise deep research tasks. Unlike existing …
Conference on Empirical Methods in Natural Language Processing (EMNLP),  2025.
Paper Cite Code
WebMMU: A Benchmark for Multimodal Multilingual Website Understanding and Code Generation
WebMMU: A Benchmark for Multimodal Multilingual Website Understanding and Code Generation
Understanding diverse web data and automating web development presents an exciting challenge for agentic multimodal models. While …
Conference on Empirical Methods in Natural Language Processing (EMNLP),  2025.
Paper Cite Code Video
GEOBench-VLM: Benchmarking Vision-Language Models for Geospatial Tasks
GEOBench-VLM: Benchmarking Vision-Language Models for Geospatial Tasks
While numerous recent benchmarks focus on evaluating generic Vision-Language Models (VLMs), they fall short in addressing the unique …
International Conference on Computer Vision (ICCV),  2025.
Paper Cite Code
BigCharts-R1: Enhanced Chart Reasoning with Visual Reinforcement Finetuning
BigCharts-R1: Enhanced Chart Reasoning with Visual Reinforcement Finetuning
Charts are essential to data analysis, transforming raw data into clear visual representations that support human decision-making. …
Conference on Language Modeling (COLM),  2025.
Paper Cite Code Video