Instructions to use fancyfeast/llama-joycaption-beta-one-hf-llava with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fancyfeast/llama-joycaption-beta-one-hf-llava with Transformers:

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

pipe = pipeline("image-text-to-text", model="fancyfeast/llama-joycaption-beta-one-hf-llava")
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("fancyfeast/llama-joycaption-beta-one-hf-llava")
model = AutoModelForMultimodalLM.from_pretrained("fancyfeast/llama-joycaption-beta-one-hf-llava")
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 fancyfeast/llama-joycaption-beta-one-hf-llava with vLLM:

Install from pip and serve model

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

SGLang

How to use fancyfeast/llama-joycaption-beta-one-hf-llava 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 "fancyfeast/llama-joycaption-beta-one-hf-llava" \
    --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": "fancyfeast/llama-joycaption-beta-one-hf-llava",
		"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 "fancyfeast/llama-joycaption-beta-one-hf-llava" \
        --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": "fancyfeast/llama-joycaption-beta-one-hf-llava",
		"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 fancyfeast/llama-joycaption-beta-one-hf-llava with Docker Model Runner:
```
docker model run hf.co/fancyfeast/llama-joycaption-beta-one-hf-llava
```

fancyfeast commited on May 12, 2025

Commit

3101114

verified ·

1 Parent(s): 537fbaf

Update README.md

Browse files

Files changed (1) hide show

README.md +4 -4

README.md CHANGED Viewed

@@ -9,10 +9,10 @@ tags:
 [Github](https://github.com/fpgaminer/joycaption)
-JoyCaption is an image captioning Visual Language Model (VLM) being built from the ground up as a free, open, and uncensored model for the community to use in training Diffusion models.
 Key Features:
-- **Free and Open**: It will be released for free, open weights, no restrictions, and just like [bigASP](https://www.reddit.com/r/StableDiffusion/comments/1dbasvx/the_gory_details_of_finetuning_sdxl_for_30m/), will come with training scripts and lots of juicy details on how it gets built.
 - **Uncensored**: Equal coverage of SFW and NSFW concepts. No "cylindrical shaped object with a white substance coming out on it" here.
 - **Diversity**: All are welcome here. Do you like digital art? Photoreal? Anime? Furry? JoyCaption is for everyone. Pains are being taken to ensure broad coverage of image styles, content, ethnicity, gender, orientation, etc.
 - **Minimal Filtering**: JoyCaption is trained on large swathes of images so that it can understand almost all aspects of our world. almost. Illegal content will never be tolerated in JoyCaption's training.
@@ -39,7 +39,7 @@ from transformers import AutoProcessor, LlavaForConditionalGeneration
 IMAGE_PATH = "image.jpg"
 PROMPT = "Write a long descriptive caption for this image in a formal tone."
-MODEL_NAME = "fancyfeast/llama-joycaption-alpha-two-hf-llava"
 # Load JoyCaption
@@ -79,7 +79,7 @@ with torch.no_grad():
 	# Generate the captions
 	generate_ids = llava_model.generate(
 		**inputs,
-		max_new_tokens=300,
 		do_sample=True,
 		suppress_tokens=None,
 		use_cache=True,

 [Github](https://github.com/fpgaminer/joycaption)
+JoyCaption is an image captioning Visual Language Model (VLM) built from the ground up as a free, open, and uncensored model for the community to use in training Diffusion models.
 Key Features:
+- **Free and Open**: Always released for free, open weights, no restrictions, and just like [bigASP](https://www.reddit.com/r/StableDiffusion/comments/1dbasvx/the_gory_details_of_finetuning_sdxl_for_30m/), will come with training scripts and lots of juicy details on how it gets built.
 - **Uncensored**: Equal coverage of SFW and NSFW concepts. No "cylindrical shaped object with a white substance coming out on it" here.
 - **Diversity**: All are welcome here. Do you like digital art? Photoreal? Anime? Furry? JoyCaption is for everyone. Pains are being taken to ensure broad coverage of image styles, content, ethnicity, gender, orientation, etc.
 - **Minimal Filtering**: JoyCaption is trained on large swathes of images so that it can understand almost all aspects of our world. almost. Illegal content will never be tolerated in JoyCaption's training.
 IMAGE_PATH = "image.jpg"
 PROMPT = "Write a long descriptive caption for this image in a formal tone."
+MODEL_NAME = "fancyfeast/llama-joycaption-beta-one-hf-llava"
 # Load JoyCaption
 	# Generate the captions
 	generate_ids = llava_model.generate(
 		**inputs,
+		max_new_tokens=512,
 		do_sample=True,
 		suppress_tokens=None,
 		use_cache=True,