farbodtavakkoli/OTel-LLM-1B-IT

OTel-LLM-1B-IT: Context-Grounded Telecom Language Model

OTel-LLM-1B-IT is a 1 billion parameter language model developed by farbodtavakkoli, specifically fine-tuned for the telecommunications sector. Built upon the google/gemma-3-1b-it base model, it is a key component of the OTel Family of Models, an open-source initiative dedicated to creating AI resources for global telecommunications. This model excels in context-grounded answer generation within Retrieval-Augmented Generation (RAG) pipelines, utilizing a 32768 token context length.

Key Capabilities

• Specialized Telecom Knowledge: Full-parameter fine-tuned on extensive OTel telecommunications data, curated by over 100 domain experts.
• Enhanced Correctness: Demonstrates an 8.5 percentage point improvement in LLM-as-judge correctness on held-out OTel evaluation data compared to its base model.
• RAG Optimization: Designed for precise, context-grounded answer generation when provided with retrieved telecom context.
• Robust Training: Trained using ScalarLM framework with AdamW optimizer, BF16 precision, and Flash Attention 2 on AMD MI300X/MI325X/MI355X and NVIDIA A100/H100 GPUs.

Good For

• Telecom RAG Systems: Ideal for applications requiring accurate, context-grounded answers to telecommunications questions.
• Domain-Specific QA: Use cases where precise information retrieval and generation within the telecom domain are critical.
• Research and Development: As a reference AI resource for the telecommunications sector, supporting further innovation and evaluation.

Note: This model is domain-specific and English-only. It is not optimized for unrestricted, context-free general question answering.