farbodtavakkoli/OTel-LLM-E4B-IT
OTel-LLM-E4B-IT is a 7.9 billion parameter context-grounded telecom language model developed by farbodtavakkoli, fine-tuned on OTel telecommunications data. Based on google/gemma-4-E4B-it, this model is specifically optimized for accurate answer generation within Retrieval-Augmented Generation (RAG) pipelines for the telecommunications sector. It demonstrates significant improvements in context-grounded correctness, achieving 91.7% on OTel-LLM held-out evaluations, making it ideal for specialized telecom QA applications.
Loading preview...
OTel-LLM-E4B-IT: A Specialized Telecom Language Model
OTel-LLM-E4B-IT is a 7.9 billion parameter language model developed by farbodtavakkoli, specifically fine-tuned on a curated dataset of telecommunications data. Part of the OTel Family of Models, this model is designed to provide context-grounded answers within the telecom domain.
Key Capabilities & Differentiators
- Domain-Specific Accuracy: Achieves 91.7% context-grounded correctness on OTel-LLM held-out evaluations, representing a +9.3 percentage point improvement over its base model,
google/gemma-4-E4B-it. - Specialized Training Data: Fine-tuned on a unique dataset curated by over 100 domain experts, including arXiv telecom papers, 3GPP standards, GSMA documents, and O-RAN specifications.
- RAG Optimization: Intended for use in Retrieval-Augmented Generation (RAG) pipelines, where it generates answers grounded in provided telecom context.
- Open-Source Initiative: Contributes to the Open Telco AI project, aiming to build reference AI resources for the global telecommunications sector.
Intended Use Cases
- Context-grounded telecom answer generation: Ideal for applications requiring precise answers based on retrieved telecommunications documents.
- RAG pipelines: Designed to enhance the accuracy of information retrieval and synthesis in telecom-specific RAG systems.
Limitations
- Domain-Specific: Not intended as a general-purpose language model; performance is optimized for telecommunications.
- English-only: The current release supports only the English language.
- Context-Dependent: Performance relies on the quality and sufficiency of the retrieved context; not optimized for unrestricted context-free question answering.