northenlab/soilfm-qwen2.5-14b-literature-cpt
TEXT GENERATIONConcurrency Cost:1Model Size:14.8BQuant:FP8Ctx Length:32kPublished:Dec 31, 2025License:apache-2.0Architecture:Transformer Open Weights Cold

The northenlab/soilfm-qwen2.5-14b-literature-cpt is a 14.2 billion parameter Qwen2.5-based causal language model, domain-adapted for soil science and soil microbiology. Developed by Northen Lab at Lawrence Berkeley National Laboratory, it was created by continued pretraining on 200,000 curated soil science text passages. This model excels at generating explanations of soil microbial processes and serves as a soil-science-aware backbone for research and educational applications.

Loading preview...