Instructions to use icdevelop/pai1-medgemma-27b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use icdevelop/pai1-medgemma-27b with Transformers:

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

pipe = pipeline("image-text-to-text", model="icdevelop/pai1-medgemma-27b")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("icdevelop/pai1-medgemma-27b")
model = AutoModelForMultimodalLM.from_pretrained("icdevelop/pai1-medgemma-27b", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use icdevelop/pai1-medgemma-27b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "icdevelop/pai1-medgemma-27b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "icdevelop/pai1-medgemma-27b",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/icdevelop/pai1-medgemma-27b

SGLang

How to use icdevelop/pai1-medgemma-27b 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 "icdevelop/pai1-medgemma-27b" \
    --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": "icdevelop/pai1-medgemma-27b",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "icdevelop/pai1-medgemma-27b" \
        --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": "icdevelop/pai1-medgemma-27b",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use icdevelop/pai1-medgemma-27b with Docker Model Runner:
```
docker model run hf.co/icdevelop/pai1-medgemma-27b
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

ต้องอ่านและยอมรับก่อนดาวน์โหลด / You must read and accept before downloading.

โมเดลและข้อมูลนี้ ไม่ใช่เครื่องมือแพทย์ ไม่ใช่แพทย์หรือเภสัชกร และไม่ใช่คำแนะนำทางการแพทย์ ข้อมูลบางส่วนสร้างโดย AI และ ยังไม่ได้รับการตรวจสอบโดยเภสัชกร (อยู่ระหว่างการตรวจสอบ) ผู้ใช้ต้องให้ผู้เชี่ยวชาญตรวจสอบข้อมูลก่อนนำไปใช้

This model/data is not a medical device, not a doctor or pharmacist, and not medical advice. By accepting you acknowledge: (1) not for diagnosis/treatment — every output must be reviewed by a licensed pharmacist; (2) part of the KB is AI-generated and not yet pharmacist-reviewed (review in progress); (3) you must have information verified by a qualified professional before use; (4) you accept the risk, developer accepts no liability; (5) bound by HAI-DEF Terms; (6) clinical use on real patients without a supervising pharmacist/physician is prohibited.

pai1-medgemma-27b - Thai Pharmacy Assistant (v24, premium tier)

WARNING: DRAFT - pharmacist & legal review pending; drug KB not yet pharmacist-reviewed (review in progress). Released under an acknowledge-before-download gate. Verify with a licensed pharmacist before real use.

Premium-tier Thai community-pharmacy assistant (27B) fine-tuned from google/medgemma-27b-it by I C Develop Co., Ltd. v24 adds the improved deliberation training (deliberate_v2, 2 epochs, rank 32) - reasons like a clinician internally, presents as a pharmacy assistant externally. Best used with RAG (icdevelop/thai-pharma-kb). A smaller FP8 build is at icdevelop/pai1-medgemma-27b-fp8.

Benchmarks (deterministic, seed 0; pass = judge-pass AND no safety violation)

Task	v20	v24
lookup (no RAG)	46.5%	43.0%
lookup + RAG	64.5%	65.3%
deliberation	29.2%	37.7%

v24 lifts deliberation (29.2 -> 37.7) via the deliberate_v2 training, with lookup/RAG roughly held.

How it was trained

Base: google/medgemma-27b-it (Gemma-3 multimodal). LoRA rank 32, vision tower + projector frozen, 2 epochs.
Corpus thai_pharma_v24 (~33k): curated Thai pharmacy dialogues + the deliberation set (deliberate_v2: rules out contraindicated drugs with reasons, asks history, refers on red flags). Dedup + decontaminated against every eval set.
Format: BF16 full-precision merged weights (~52 GB).

How to use (RAG recommended)

vllm serve icdevelop/pai1-medgemma-27b --dtype bfloat16 --max-model-len 8192 --served-model-name pai --seed 0 --enforce-eager
huggingface-cli download icdevelop/thai-pharma-kb --repo-type dataset

Retrieve real KB facts and inject them into the prompt (Thai). Ground with RAG for best accuracy.

Data / limits

KB: TMT registry (~31k) + ingredient KB (1,918, 534 flagged, not pharmacist-reviewed). Knowledge cutoff July 2026. Not a medical device; not for clinical use without a licensed pharmacist.