Shinzmann/naija-petro-8b

TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Mar 20, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Shinzmann/naija-petro-8b is an 8 billion parameter Qwen3-based instruction-tuned causal language model developed by the Naija-Petro project. Fine-tuned using QLoRA on approximately 20,000 synthetic petroleum-engineering instruction-response pairs, this model is optimized for technical question answering and explanation across various petroleum subdomains. It serves as a lightweight, fast-inference variant designed for study aids, engineering decision support, and as a backbone for retrieval-augmented systems in the petroleum industry.

Loading preview...

Naija-Petro 8B: Specialized Petroleum Engineering LLM

Naija-Petro 8B is an 8 billion parameter, decoder-only causal language model, instruction-tuned from the Qwen3-8B architecture. Developed by the Naija-Petro project, this model is specifically fine-tuned using QLoRA and Unsloth on a dataset of approximately 20,000 synthetic petroleum-engineering instruction-response pairs.

Key Capabilities & Features

  • Domain Expertise: Excels in technical question answering and explanations across core petroleum engineering subdomains, including drilling, reservoir, production, completions, EOR, well testing, and petroleum geoscience.
  • Instruction-Tuned: Designed to provide precise, technically accurate answers, often including equations, units, and practical considerations, making it suitable as a study aid or engineering decision-support tool.
  • Lightweight & Fast: Positioned as a lightweight, fast-inference variant within the Naija-Petro family, ideal for integration into retrieval-augmented generation (RAG) systems.
  • Training Data: Fine-tuned on synthetic data generated from a comprehensive scraped corpus including arXiv, Semantic Scholar, OpenAlex, Crossref, DOE/OSTI, PetroWiki, SLB glossary, and EIA.

Use Cases & Limitations

Good for:

  • Direct technical question answering and explanation in petroleum engineering.
  • Serving as a backbone for retrieval-augmented assistants, especially when paired with the Naija-Petro RAG system for Nigeria-specific facts.
  • Further domain-specific fine-tuning or model distillation.

Important Considerations:

  • The model's knowledge is primarily based on general/global petroleum knowledge. For Nigeria-specific regulations or economics, it should be used with a RAG system.
  • Trained on synthetic data, it may occasionally "hallucinate" or be confidently wrong, particularly on numerical specifics. Outputs should always be validated by qualified engineers and primary sources.
  • It is not intended for autonomous operational, safety-critical, or financial decisions, nor as a substitute for licensed engineering judgment.