Instructions to use nevermindai/carejev-omni with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nevermindai/carejev-omni with Transformers:

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

pipe = pipeline("image-text-to-text", model="nevermindai/carejev-omni")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("nevermindai/carejev-omni")
model = AutoModelForMultimodalLM.from_pretrained("nevermindai/carejev-omni", 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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nevermindai/carejev-omni with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nevermindai/carejev-omni"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nevermindai/carejev-omni",
		"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/nevermindai/carejev-omni

SGLang

How to use nevermindai/carejev-omni 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 "nevermindai/carejev-omni" \
    --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": "nevermindai/carejev-omni",
		"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 "nevermindai/carejev-omni" \
        --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": "nevermindai/carejev-omni",
		"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 nevermindai/carejev-omni with Docker Model Runner:
```
docker model run hf.co/nevermindai/carejev-omni
```

carejev-omni / example.py

nevermindai

CareJev-Omni release

19d52b2 verified 18 days ago

Raw History Blame Contribute Delete

3.21 kB

"""Runnable example shipped inside the export (copied next to the loader shim as ``example.py``).
    pip install -r requirements.txt
    python example.py            # uses this directory as the checkpoint
Prints one answer of each question type. Needs a GPU with room for the checkpoint; pass a directory
as the first argument to score a different export.
"""

import sys
from pathlib import Path

HERE = Path(__file__).resolve().parent
sys.path.insert(0, str(HERE))

from carejev_omni_loader import load_carejev_omni  # noqa: E402  (shipped next to this file)

STATE = (
    "PATIENT: 68-year-old male. Race: white.\n"
    "STUDY ENTRY: hospital day 2; primary disease category: acute respiratory failure or multiple "
    "organ system failure with sepsis (ARF/MOSF).\n"
    "COMORBIDITIES: 2 recorded (diabetes).\n"
    "SEVERITY: SUPPORT coma score 0; APACHE III physiology score 61; SUPPORT physiology score 33.9.\n"
    "FUNCTION: 1 activities of daily living impaired (0 = independent, 7 = fully dependent).\n"
    "DAY-3 PHYSIOLOGY: mean arterial pressure 72 mmHg; heart rate 104 /min; respiratory rate 28 /min; "
    "temperature 38.1 °C; white cell count 15.2 x10^9/L; creatinine 2.1 mg/dL; sodium 139 mmol/L; "
    "arterial pH 7.33; blood urea nitrogen 44 mg/dL; urine output 900 mL/day."
)

LOS_LEVELS = [
    "Less than 1 day",
    "1 day",
    "2 days",
    "3 days",
    "4 days",
    "5 days",
    "6 days",
    "7 days",
    "8 to 14 days",
    "More than 14 days",
]


def main() -> None:
    path = sys.argv[1] if len(sys.argv) > 1 else str(HERE)
    clf = load_carejev_omni(path)

    # noul: a yes/no question. The answer is P(yes); option order never changes it.
    noul = clf.predict(
        state=STATE,
        question="Will this patient die before being discharged from this hospitalization?",
        options=["Yes", "No"],
    )
    print("noul   P(yes) =", round(noul["probabilities"]["Yes"], 3))

    # choice: one of k options. Probabilities are averaged over rotations of a canonical order.
    choice = clf.predict(
        state=STATE,
        question="Which single problem best explains this presentation?",
        options=["Sepsis", "Heart failure", "Pulmonary embolism", "Diabetic ketoacidosis"],
    )
    print("choice      ->", choice["prediction"], "| confidence", round(choice["confidence"], 3))

    # score: ordered levels. `score` is the expected level, i.e. sum(index * probability).
    score = clf.predict(
        state=STATE,
        question="How long will this patient's hospital stay be in total?",
        options=LOS_LEVELS,
    )
    top = max(score["probabilities"], key=score["probabilities"].get)
    expected = sum(i * p for i, p in enumerate(score["probabilities"].values()))
    print(f"score       -> {top} | expected level {expected:.2f} of {len(LOS_LEVELS) - 1}")

    # Images, heart sounds and other media go in the same call:
    #   clf.predict(state=..., question=..., options=[...], media="ecg.png", modality="image")
    #   clf.predict(state=..., question=..., options=["Yes", "No"], media="pcg.wav", modality="audio")
    print("\nResearch use only. Not a medical device. See README.md.")


if __name__ == "__main__":
    main()

	"""Runnable example shipped inside the export (copied next to the loader shim as ``example.py``).

	pip install -r requirements.txt
	python example.py # uses this directory as the checkpoint

	Prints one answer of each question type. Needs a GPU with room for the checkpoint; pass a directory
	as the first argument to score a different export.
	"""

	import sys
	from pathlib import Path

	HERE = Path(__file__).resolve().parent
	sys.path.insert(0, str(HERE))

	from carejev_omni_loader import load_carejev_omni # noqa: E402 (shipped next to this file)

	STATE = (
	"PATIENT: 68-year-old male. Race: white.\n"
	"STUDY ENTRY: hospital day 2; primary disease category: acute respiratory failure or multiple "
	"organ system failure with sepsis (ARF/MOSF).\n"
	"COMORBIDITIES: 2 recorded (diabetes).\n"
	"SEVERITY: SUPPORT coma score 0; APACHE III physiology score 61; SUPPORT physiology score 33.9.\n"
	"FUNCTION: 1 activities of daily living impaired (0 = independent, 7 = fully dependent).\n"
	"DAY-3 PHYSIOLOGY: mean arterial pressure 72 mmHg; heart rate 104 /min; respiratory rate 28 /min; "
	"temperature 38.1 °C; white cell count 15.2 x10^9/L; creatinine 2.1 mg/dL; sodium 139 mmol/L; "
	"arterial pH 7.33; blood urea nitrogen 44 mg/dL; urine output 900 mL/day."
	)

	LOS_LEVELS = [
	"Less than 1 day",
	"1 day",
	"2 days",
	"3 days",
	"4 days",
	"5 days",
	"6 days",
	"7 days",
	"8 to 14 days",
	"More than 14 days",
	]


	def main() -> None:
	path = sys.argv[1] if len(sys.argv) > 1 else str(HERE)
	clf = load_carejev_omni(path)

	# noul: a yes/no question. The answer is P(yes); option order never changes it.
	noul = clf.predict(
	state=STATE,
	question="Will this patient die before being discharged from this hospitalization?",
	options=["Yes", "No"],
	)
	print("noul P(yes) =", round(noul["probabilities"]["Yes"], 3))

	# choice: one of k options. Probabilities are averaged over rotations of a canonical order.
	choice = clf.predict(
	state=STATE,
	question="Which single problem best explains this presentation?",
	options=["Sepsis", "Heart failure", "Pulmonary embolism", "Diabetic ketoacidosis"],
	)
	print("choice ->", choice["prediction"], "\| confidence", round(choice["confidence"], 3))

	# score: ordered levels. `score` is the expected level, i.e. sum(index * probability).
	score = clf.predict(
	state=STATE,
	question="How long will this patient's hospital stay be in total?",
	options=LOS_LEVELS,
	)
	top = max(score["probabilities"], key=score["probabilities"].get)
	expected = sum(i * p for i, p in enumerate(score["probabilities"].values()))
	print(f"score -> {top} \| expected level {expected:.2f} of {len(LOS_LEVELS) - 1}")

	# Images, heart sounds and other media go in the same call:
	# clf.predict(state=..., question=..., options=[...], media="ecg.png", modality="image")
	# clf.predict(state=..., question=..., options=["Yes", "No"], media="pcg.wav", modality="audio")
	print("\nResearch use only. Not a medical device. See README.md.")


	if __name__ == "__main__":
	main()