Problem
Research questions and friction points this paper is trying to address.
geological reasoning
large language models
hallucination
subsurface structures
deep-time evolution
Geo-Expert: Towards Expert-Level Geological Reasoning via Parameter-Efficient Fine-Tuning
🤖 AI Summary
This work addresses the limitations of general-purpose large language models (LLMs), which frequently generate hallucinations regarding subsurface structures and deep-time geological evolution, while existing Earth science AI systems remain largely confined to surface remote sensing and lack specialized geological reasoning capabilities. To bridge this gap, the authors construct a high-quality geological instruction dataset and synthesis pipeline, then apply domain-aligned, parameter-efficient fine-tuning using LoRA on Qwen3-8B/32B and Gemma-3-27B. The resulting 8B model outperforms both 70B-class open-source LLMs and GPT-4o on the Geo-Eval benchmark, while the 32B variant approaches state-of-the-art reasoning performance—demonstrating an optimal balance between accuracy and deployment cost. This study establishes a reproducible paradigm for building expert-level scientific foundation models tailored to geological reasoning.
📝 Abstract
While general-purpose Large Language Models (LLMs) applied to Geology often hallucinate when reasoning about subsurface structures and deep-time evolution, current AI in Earth sciences predominantly targets surface remote sensing and GIS. To bridge this gap, we introduce Geo-Expert, a family of parameter-efficient geological LLMs fine-tuned on a custom-curated, high-quality instruction dataset processed using our custom instruction synthesis pipeline. We investigate the impact of model scaling and architecture by fine-tuning three base models: Qwen3-8B, Qwen3-32B, and Gemma-3-27B, with Low-Rank Adaptation (LoRA) method. Our extensive evaluation on a novel domain-specific benchmark, Geo-Eval, reveals that a domain-aligned 8B model can outperform open-weight 70B generalists and proprietary GPT-4o on specialized geological reasoning, while a 32B variant approaches frontier reasoning models. The optimized 8B model further offers a competitive cost-performance ratio for deployment. This work provides a reproducible recipe for democratizing scientific LLMs and establishes a baseline for geological artificial intelligence.
Innovation
Methods, ideas, or system contributions that make the work stand out.
parameter-efficient fine-tuning
geological reasoning
instruction synthesis
LoRA
domain-specific LLM
🔎 Similar Papers
No similar papers found.
