esa-sceva/llama3-satcom-70b

Overview

esa-sceva/llama3-satcom-70b is a 70 billion parameter, instruction fine-tuned Large Language Model (LLM) based on meta-llama/Llama-3.3-70B-Instruct. Developed under the ESA ARTES programme as part of the SatcomLLM / SCEVA project, it is the largest and most capable SatCom-specialised LLM. It is designed to support satellite communications (SatCom) professionals with advanced reasoning, multi-step problem solving, and expert-level Q&A.

Key Capabilities & Features

• Domain Specialization: Purpose-built for satellite communications, offering enhanced comprehension of technical systems and contextual understanding across complex SatCom documents.
• Advanced Reasoning: Excels in tasks such as link budget evaluation, propagation modeling, 5G/6G NTN design, and mission architecture analysis.
• Training Data: Fine-tuned on curated and synthetic SatCom QA, including chain-of-thought annotated datasets (esa-sceva/satcom-synth-qa, esa-sceva/satcom-synth-qa-cot) to strengthen reasoning accuracy.
• Context Length: Features an extended context window of 128k tokens, enabling processing of longer technical passages and integration of multiple data sources.
• Performance: Achieves strong results on SatCom-specific benchmarks, outperforming its base model in MCQA, Satcom-QA, and EVE-QA evaluations.

Intended Use Cases

• Advanced reasoning and Q&A for SatCom system design and analysis.
• Automated support for link budget calculations and RF engineering tasks.
• Conceptual guidance for 5G/6G NTN and inter-satellite network operations.
• Mission design evaluation and anomaly diagnosis assistance.
• Research, education, and technical documentation in satellite communications.

Limitations

• Does not access live telemetry or proprietary ESA mission data.
• Generated answers require validation by domain experts before operational use.
• Not suitable for safety-critical or real-time decision-making.
• May produce confident but incorrect answers outside its specialized domain.