farbodtavakkoli/OTel-LLM-270M-IT

OTel-LLM-270M-IT: A Specialized Telecom Language Model

OTel-LLM-270M-IT is a 270 million parameter instruction-tuned language model developed by farbodtavakkoli, specifically designed for the telecommunications sector. It is a key component of the OTel Family of Models, an open-source initiative focused on creating industry-standard AI for global telecom. This model is built on the google/gemma-3-270m-it base and was fine-tuned using a comprehensive dataset of telecom-focused information, including arXiv papers, 3GPP standards, GSMA documents, IETF RFCs, and O-RAN specifications, contributed by various institutional partners.

Key Capabilities

• Telecom Domain Expertise: Fine-tuned on data curated by over 100 domain experts, ensuring deep understanding of telecommunications concepts and terminology.
• Retrieval-Augmented Generation (RAG) Optimized: Designed to work within RAG pipelines, complementing OTel Embedding and Reranker models for end-to-end information retrieval and generation.
• Hallucination Prevention: Incorporates abstention training, allowing the model to decline answering if insufficient context is provided, thereby enhancing factual accuracy.
• Apache 2.0 Licensed: Available for broad use under an open-source license.

Good for

• Building Telecom RAG Systems: Ideal for developers creating applications that require accurate, context-grounded responses from telecommunications documentation.
• Answering Telecom-Specific Queries: Generating precise answers to questions based on industry standards, specifications, and research papers.
• Integrating with OTel Ecosystem: Seamlessly combines with other OTel models (Embedding, Reranker) to form a robust AI solution for the telecom industry.