Instructions to use nghiaump/mda2-v23 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nghiaump/mda2-v23 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="nghiaump/mda2-v23")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nghiaump/mda2-v23")
model = AutoModelForCausalLM.from_pretrained("nghiaump/mda2-v23", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nghiaump/mda2-v23 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nghiaump/mda2-v23"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nghiaump/mda2-v23",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nghiaump/mda2-v23

SGLang

How to use nghiaump/mda2-v23 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "nghiaump/mda2-v23" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nghiaump/mda2-v23",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "nghiaump/mda2-v23" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nghiaump/mda2-v23",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nghiaump/mda2-v23 with Docker Model Runner:
```
docker model run hf.co/nghiaump/mda2-v23
```

MDA2-V23: Vietnamese Medical History-Taking Model

Fine-tuned Qwen3-4B-Instruct for Vietnamese medical history-taking (khai thac benh su).

Architecture

Python Orchestrator (Master) + LLM (Executor) - The model handles 6 NLP tasks while Python controls clinical logic.

Training

Parameter	Value
Base model	Qwen/Qwen3-4B-Instruct-2507
Method	QLoRA (r=64, alpha=128, NF4)
Training data	52,000 examples (6 tasks)
Epochs	3
Batch size	8 (grad_accum=12, effective=96)
Learning rate	2e-4
Seq length	1536
Hardware	H100 80GB MIG (4g.40gb)
Duration	7h
Final avg loss	0.187

6 Tasks

Task	Count	Description
T1 NLG	12,000	Generate natural Vietnamese doctor questions
T2 Diagnosis	4,000	Clinical reasoning + differential diagnosis
T3 Extraction	8,000	Extract symptoms from patient speech
T4 Characterization	10,000	Extract OPQRST field values
T5 Summary	10,000	Medical history summarization
T6 Safety	8,000	Red flag detection + urgency assessment

Key Features (V23 improvements)

Think blocks: All tasks output for chain-of-thought
KB-driven data: Real OPQRST from 1,227 edges, 870 red flags from 357 symptoms
8 clinical frameworks: OLDCARTS, DYSPNEA, GI, EDEMA, CONSTITUTIONAL, URINARY, SKIN, SIMPLE
Claude API generation: T2 diagnosis uses Haiku for realistic clinical reasoning

Usage

Downloads last month: 5

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for nghiaump/mda2-v23

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(1842)

this model