Papers for

defense technology teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Benchmark tests ai on geopolitical report question answering

GeopoliticalQA: Benchmarking Retrieval and Generation over Intelligence Reports

Abstract: Geopolitical and strategic intelligence reports constitute a rich source of expert knowledge and have increasing value in the current political climate, but their heterogeneity and analytical complexity pose challenges for retrieval-augmented question answering. We introduce a benchmark for grounded question answering over 415 open-source geopolitical and strategic intelligence reports, comprising 5,525 document-grounded question-answer pairs spanning factual, inferential, comparative, and risk-oriented questions. We propose a document-aware generation pipeline that profiles reports before generating questions and validate dataset quality through expert annotation and an LLM-as-judge protocol. Using the benchmark, we evaluate 15 retrieval systems and six open-weight LLMs for RAG. We find that models perform similarly on document-level retrieval, while at the harder chunk level retrieval quality is not driven by embedding-model scale: bge-m3, a mid-size open encoder matches an order of magnitude larger ones and outperforms commercial embedding APIs. In contrast, answer generation remains challenging even given the gold evidence, particularly for inferential and risk-oriented questions.

Thu 8 OctComputation and LanguageInformation Retrieval
The gist
Understanding complex intelligence reports is tough because they are very detailed and cover many topics. The authors created a special set of questions and answers based on these reports to test how well AI systems can find information and give answers. They checked many different AI systems and found that while some can retrieve documents well, actually answering tricky questions is still hard. Their work helps show which parts of AI need improvement to better handle these important reports.
Open → 2610.12617v1

Salute improves large language models for military defense tasks

SALUTE: Benchmarking and Adapting LLMs for the Defense Domain

Abstract: Defense is a knowledge-intensive domain that requires precise understanding of specialized terminology, doctrinal concepts, operational procedures, and evolving military events. Although recent work has explored language technologies for military applications, existing efforts remain fragmented: they are often task-specific, rely on limited adaptation pipelines, or lack comprehensive defense-domain evaluation. In this paper, we present SALUTE, an end-to-end framework for benchmarking and adapting LLMs for the defense domain. SALUTE integrates Salute-Corpus, a curated corpus from open-access U.S. military doctrine and government documents; Salute-Conv, a grounded instruction dataset from doctrinal sources and decade-long defense news; Salute-Pref, a defense-aware preference dataset; and Salute-Bench, a rigorously filtered benchmark for evaluating defense-domain understanding and reasoning over doctrine and defense news. Based on these resources, we train Salute-LLM through multi-stage post-training with continual pretraining, supervised fine-tuning, and preference alignment. Extensive experiments show that Salute-LLM achieves strong defense-domain performance while retaining competitive general capabilities, demonstrating the effectiveness of SALUTE as an end-to-end framework for defense-domain LLM adaptation.

Mon 14 SeptComputation and Language
The gist
Understanding military language and concepts is difficult because they use special words and ideas that change often. The authors created SALUTE, a complete approach to help large language models learn and understand defense language better by using military documents and news. They also made tests to check how well the models understand military topics. Their improved model, Salute-LLM, understands military texts well and still works well on everyday language tasks.
Open → 2609.15022v1