Downloads · 30 days
48
4% of all-time downloads
pnnbao-ump/MedCrab-1.5B
MedCrab-1.5B is a translation model from pnnbao-ump. Use it when you need text moved from one language to another. The card lists the license as cc-by-nc-4.0.
[](https://github.com/pnnbao97/MedCrab) [](https://huggingface.co/pnnbao-ump/MedCrab-1.5B) [](https://huggingface.co/datasets/pnnbao-ump/MedCrab)
Downloads · 30 days
48
4% of all-time downloads
All-time downloads
1.2K
Public
Parameters
1.5B
30.9 GB on disk
Likes
14
Public
Click a slice to open those files.
.safetensors3.1 GB · 99%
From the Hugging Face model README

A specialized English→Vietnamese medical translation model for clinical documents, research papers, and biomedical engineering content.
Trained on the MedCrab dataset — 300,000 English-Vietnamese medical translation pairs validated by clinical physicians both in Vietnam and internationally.
📊 Dataset: pnnbao-ump/MedCrab
<video controls width="100%"> <source src="https://cdn-uploads.huggingface.co/production/uploads/68b923a86c86c127a1975eda/ho-nDkDXD2bPSvHXbT7Bp.mp4" type="video/mp4"> Trình duyệt của bạn không hỗ trợ thẻ video. </video># Install uv
pip install uv
# Clone and setup
git clone https://github.com/pnnbao97/MedCrab.git
cd MedCrab
uv sync
# Run translator
# If your GPU has >= 8 GB VRAM, you can run PDF/Image translation:
uv run pdf_translator.py
# Otherwise, for lower VRAM, run main translator:
uv run main.py
Evaluated on complex medical passages (100-150 words) covering multi-omics, cellular biology, and pathology:
| Metric | Score |
|---|---|
| BLEU | 42–46 |
| COMET | 0.68–0.72 |
| METEOR | 37–40 |
Key strengths: High fidelity preservation of biomarkers, pathways, and technical terminology with publication-ready fluency. Maintains stability on 100+ word passages where smaller models (<0.5B) degrade.
Input:
Recent integrative analyses combining single-cell RNA sequencing, spatial transcriptomics, and high-dimensional mass cytometry have identified a previously uncharacterized population of CD141⁺ dendritic cells in the fibrotic niche of patients with non-alcoholic steatohepatitis (NASH).
Output:
Phân tích tích hợp gần đây kết hợp giải trình tự RNA đơn bào, phiên mã không gian, và đo khối tế bào đa chiều đã xác định quần thể tế bào tua CD141⁺ chưa rõ đặc điểm ở ổ xơ hóa của bệnh nhân viêm gan nhiễm mỡ không do rượu (NASH).
Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
This project is licensed under CC BY-NC 4.0. You are free to:
Commercial Use: All commercial applications require direct permission from the author.
For research purposes only. Not intended for medical diagnosis or treatment decisions.
Author: Phạm Nguyễn Ngọc Bảo | Facebook