esa-sceva/llama3-satcom-8b

Model Overview

esa-sceva/llama3-satcom-8b is an 8 billion parameter instruction-tuned Large Language Model (LLM) based on meta-llama/Llama-3.1-8B-Instruct. Developed under the ESA ARTES programme as part of the SatcomLLM / SCEVA project, this model is specifically fine-tuned to assist satellite communications (SatCom) experts, engineers, and mission planners.

Key Capabilities & Differentiators

• Domain-Specific Expertise: Fine-tuned on curated SatCom-related corpora, including synthetic QA data with Chain-of-Thought (CoT) annotations, to enhance understanding of technical language, protocols, and reasoning processes unique to satellite communications.
• Specialized Reasoning: Excels in areas such as 5G/6G non-terrestrial networks (NTN), link budget evaluation, and mission engineering tasks.
• Instruction Fine-Tuning (IFT): Utilizes instruction fine-tuning on domain-specific question-answer datasets to improve its ability to respond to technical queries.
• Performance: Achieves improved accuracy on SatCom-specific multiple-choice and open-ended QA tasks compared to its base model, Llama-3.1-8B-Instruct, demonstrating enhanced domain understanding.

Intended Use Cases

• Technical Q&A: Answering questions and providing reasoning on SatCom systems, RF engineering, and 5G/6G NTN operations.
• Mission Support: Assisting with mission design, planning, and anomaly detection within the SatCom sector.
• Educational & Research: Supporting academic and research activities for SatCom and aerospace professionals.

Limitations

It's important to note that the model does not access real-time mission data or proprietary ESA documents. Its answers are based on training data and require expert validation for operational use, and it should not be relied upon for flight-critical or safety-critical decisions.