Model Card for Lingjian-1.0-Chat

Lingjian-1.0-Chat is a specialized vision-language model for clinical laboratory medicine, achieving comprehensive capabilities in microscopic image understanding, cell identification, morphological analysis, and target localization.

Model Details

Model Description

Lingjian-1.0-Chat ("灵检" signifies "Intelligent Laboratory Testing" in Chinese) is a domain-adapted vision-language model specifically designed for clinical laboratory applications. Through multi-stage training on over 400,000 clinical laboratory images, the model unifies image understanding, cell identification, morphological analysis, and key target localization within a single framework.

• Developed by: LingJian AI Research Team
• Model type: Vision-Language Model for Medical Imaging
• Language(s) (NLP): Chinese (primary), English
• License: Apache-2.0
• Finetuned from model: Qwen/Qwen2.5-VL-7B-Instruct

Model Sources [optional]

• Repository: https://huggingface.co/LingJianAI/Lingjian-1.0-Chat
• Paper: Technical report in preparation
• Demo: https://huggingface.co/spaces/LingJianAI/LingJian-1.0-demo

Uses

Direct Use

• Clinical auxiliary diagnosis and secondary review
• Laboratory quality control and morphology training
• Medical education and teaching demonstrations
• Research data screening and analysis

Downstream Use [optional]

• Specialized fine-tuning for specific laboratory sub-domains
• Integration with laboratory information systems
• Development of automated reporting tools

Out-of-Scope Use

• Primary diagnosis without human supervision
• Applications outside clinical laboratory medicine
• High-precision localization tasks requiring pixel-level accuracy
• Non-medical image analysis

Bias, Risks, and Limitations

• Bounding box precision may be inferior to specialized detection models
• Performance may vary with rare cell types or unusual staining conditions
• Currently optimized for Chinese medical terminology
• Training data limitations in certain sub-specialties
• Requires validation in real clinical settings

Recommendations

Users should be aware that this model is intended as an auxiliary tool and requires verification by qualified medical professionals. Clinical decisions should not be based solely on model predictions without human oversight.

How to Get Started with the Model

Use the code below to get started with the model.

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model = AutoModelForCausalLM.from_pretrained(
    "LingJianAI/Lingjian-1.0-Chat",
    torch_dtype=torch.float16,
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("LingJianAI/Lingjian-1.0-Chat")

# For target localization, append "/bbox" to instruction
instruction = "识别图中的关键细胞并标记位置。/bbox"