Mapping the Vertical LLM Sprawl

What it does This repository is an awesome-list-style index that catalogs open-source domain-specific LLMs, along with the datasets and evaluation benchmarks used to create them. It organizes models by vertical—medical, legal, financial, educational, and others—while also listing the general base models they build upon. Each entry typically links to the code, paper, and hosting institution.

The interesting bit The list doubles as a snapshot of the Chinese LLM ecosystem’s vertical expansion, with models named after historical figures like HuaTuo and BianQue for medical tasks, plus coverage of cybersecurity, marine science, and agriculture. A pipeline diagram illustrates the standard path: take a general model, then apply continued pre-training or instruction fine-tuning for a specific field.

Key highlights

• Covers dozens of vertical models across medical, legal, financial, educational, DevOps, cybersecurity, and agricultural domains.
• Tracks underlying base models—LLaMA2, ChatGLM3, Qwen, Baichuan2, InternLM—with size and paper references.
• Includes domain-specific benchmarks such as LawBench, FinanceBench, DevOps-Eval, and CFBenchmark.
• Many entries focus on Chinese-language capabilities and domestic institutions (Fudan, Alibaba, Tsinghua).
• The visible update log runs through late 2023, suggesting a period of rapid curation.

Caveats

• The README and all annotations are in Chinese, so non-readers will need a translator.
• It is a reading list, not a framework; there is no unified API or training pipeline to run.
• The changelog in the sources stops at November 2023, so recent models may be missing.

Verdict Bookmark this if you are researching domain adaptation or hunting for a Chinese-language vertical model to fine-tune. Skip it if you are looking for a deployable software package.