farbodtavakkoli/OTel-LLM-7B-IT

OTel-LLM-7B-IT: A Specialized Telecom Language Model

OTel-LLM-7B-IT is a 7 billion parameter instruction-tuned language model developed by farbodtavakkoli, specifically designed for the telecommunications sector. It is a full-parameter fine-tuned version of the allenai/Olmo-3-7B-Instruct base model, leveraging the comprehensive OTel telecommunications dataset. This model is part of the OTel Family of Models, an open-source initiative aimed at providing reference AI resources for global telecommunications.

Key Capabilities

• Context-Grounded Telecom QA: Optimized for generating accurate answers to telecommunications questions when provided with retrieved context, making it highly effective for RAG pipelines.
• Improved Correctness: Demonstrates a significant improvement in context-grounded correctness, ranging from +3.3 to +9.4 percentage points over its base checkpoints on OTel telecommunications data.
• Specialized Training Data: Trained on a curated dataset of approximately 326,767 high-confidence telecom-focused examples, including arXiv papers, 3GPP standards, GSMA documents, and O-RAN specifications.

Good For

• Retrieval-Augmented Generation (RAG) in Telecom: Ideal for applications requiring precise, context-grounded answers to telecom-specific queries.
• Domain-Specific AI Solutions: Suitable for developers building AI tools and services tailored for the telecommunications industry.
• Research and Development: Provides a specialized foundation for further research into AI applications within the telecom domain.