Search NASA⌕ Search

SEARCH · Search NASA

Results for “Large language models”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Solving tricky quantum optics problems with assistance from large language models

The capabilities of modern artificial intelligence (AI) as a “scientific collaborator” are explored by engaging it with three nuanced problems in quantum optics: state populations in optical pumping, resonant transitions between decaying states (the Burshtein effect), and degenerate mirrorless lasing. Through iterative dialogue, the authors observe that AI models–when prompted and corrected–can reason through complex scenarios, refine their answers, and provide expert-level guidance, closely resembling the interaction with an adept colleague. The findings highlight that AI can democratize access to sophisticated modeling and analysis, shifting the focus in scientific practice from technical mastery to the generation and testing of ideas, and reducing the time for completing research tasks from days to minutes.

74 ATOMIC AND MOLECULAR PHYSICS↗
DOE OSTI2022

Towards Automatic Mapping of Vulnerabilities to Attack Patterns using Large Language Models

With the advent of new devices and applications, cyber attack surface is continuously evolving due to the emergence of new attack techniques and vulnerabilities. Hence, security management tool must assess the cyber risk of an enterprise at regular interval basis through comprehensively identifying associations among attack techniques, weakness, and vulnerabilities. However, existing repositories providing such associations are incomplete (i.e., missing associations), inducing the likelihood of undermining the risk of particular set of attack techniques. Moreover, such associations still rely on manual interpretation, which is slow compared to attack speed and ineffective for the increasing list of vulnerabilities and attack actions. Therefore, there is an urge to develop methodologies for automatically associating vulnerabilities to all relevant attack techniques. In this paper, we present a framework, named VWC-MAP, that can automatically identify all relevant attack techniques of a vulnerability via weakness based on their text descriptions, applying natural language process (NLP) techniques. To achieve that, we present a novel two-tiered classification approach, where the first tier classifies vulnerabilities to weakness, and the second tier classifies weakness to attack techniques. This research has improved the scalability of the current state-of-the-art tool to make vulnerability to weakness mapping significantly faster. Moreover, this paper presents two novel approaches for weakness to attack technique mapping applying Text-to-Text and link prediction techniques. Our experiment results cross-validated through cyber-security experts show that VWC-MAP can associate vulnerabilities to weakness types with 87% accuracy and to new attack patterns with 80% accuracy.

Das, Siddhartha Shankar↗
DOE OSTI2026

FLYING SERVING: On-the-Fly Parallelism Switching for Large Language Model Serving

Production LLM serving must simultaneously deliver high throughput, low latency, and sufficient context capacity under non-stationary traffic and mixed request requirements. Data parallelism (DP) maximizes throughput by running independent replicas, while tensor parallelism (TP) reduces per-request latency and pools memory for long-context inference. However, existing serving stacks typically commit to a static parallelism configuration at deployment; adapting to bursts, priorities, or long-context requests is often disruptive and slow. We present Flying Serving, a vLLM-based system that enables online DP-TP switching without restarting engine workers. Flying Serving makes reconfiguration practical by virtualizing the state that would otherwise force data movement: (i) a zero-copy Model Weights Manager that exposes TP shard views on demand, (ii) a KV Cache Adaptor that preserves request KV state across DP/TP layouts, (iii) an eagerly initialized Communicator Pool to amortize collective setup, and (iv) a deadlock-free scheduler that coordinates safe transitions under execution skew. Across three popular LLMs and realistic serving scenarios, Flying Serving improves performance by up to 4.79 × under high load and 3.47 × under low load while supporting latency- and memory-driven requests.

Gao, Shouwei [ORNL]↗