Instructions to use xiaokuixk/PortraitCraft-Track1-sky with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use xiaokuixk/PortraitCraft-Track1-sky with Transformers:

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

pipe = pipeline("image-text-to-text", model="xiaokuixk/PortraitCraft-Track1-sky")
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("xiaokuixk/PortraitCraft-Track1-sky")
model = AutoModelForMultimodalLM.from_pretrained("xiaokuixk/PortraitCraft-Track1-sky")
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 xiaokuixk/PortraitCraft-Track1-sky with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "xiaokuixk/PortraitCraft-Track1-sky"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "xiaokuixk/PortraitCraft-Track1-sky",
		"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/xiaokuixk/PortraitCraft-Track1-sky

SGLang

How to use xiaokuixk/PortraitCraft-Track1-sky 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 "xiaokuixk/PortraitCraft-Track1-sky" \
    --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": "xiaokuixk/PortraitCraft-Track1-sky",
		"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 "xiaokuixk/PortraitCraft-Track1-sky" \
        --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": "xiaokuixk/PortraitCraft-Track1-sky",
		"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 xiaokuixk/PortraitCraft-Track1-sky with Docker Model Runner:
```
docker model run hf.co/xiaokuixk/PortraitCraft-Track1-sky
```

xiaokuixk commited on 28 days ago

Commit

04179ee

verified ·

1 Parent(s): 8cb0e75

Update README.md

Browse files

Files changed (1) hide show

README.md +1 -27

README.md CHANGED Viewed

@@ -31,10 +31,6 @@ All three are emitted in a single strict-JSON output.
 ## Quick start
 ```bash
-git lfs install
-git clone https://huggingface.co/<your-namespace>/<your-repo>
-cd <your-repo>
 pip install -r inference/requirements.txt
 # Run inference on the official Track-1 test set:
@@ -45,16 +41,6 @@ bash run_inference.sh \
 This produces `submission.json` and `submission.zip` in the repo root.
-## Model details
-- **Base model**: Qwen3.5-4B-VL (multimodal variant)
-- **Adapter**: LoRA (rank=64, alpha=128, dropout=0.05, all linear projection layers)
-- **Training**: 20 epochs, constant learning rate 2e-5, effective batch size 16
-- **Precision**: bfloat16
-- **Hardware**: 4 × NVIDIA H20 GPUs
-- **Distillation**: trained against soft labels averaged from a multi-model teacher ensemble on the official training set
-- **Adapter merge**: LoRA weights merged into the base for single-engine vLLM serving (8.5 GB safetensors)
 ## Inference
 The pipeline at inference time uses **2-pass test-time augmentation**:
@@ -93,18 +79,6 @@ The continuous criterion scores from all four passes are averaged, and only then
 }
 ```
-## Repository contents
-| File | Description |
-|------|-------------|
-| `model.safetensors` | Merged model weights (Qwen3.5-4B-VL + LoRA, bf16, 8.5 GB) |
-| `config.json` + tokenizer files | Standard HF model artefacts |
-| `chat_template.jinja` | Chat template used at training/inference time |
-| `inference/infer_score.py` | Single-resolution + flip-TTA inference |
-| `inference/make_submission.py` | Two-pass merge + score→level + final submission JSON |
-| `inference/requirements.txt` | Pinned dependencies |
-| `run_inference.sh` | One-command end-to-end driver |
 ## Environment
 Pinned versions for reproducibility (see `inference/requirements.txt`):
@@ -116,7 +90,7 @@ Pinned versions for reproducibility (see `inference/requirements.txt`):
 | torch | 2.10.0 (CUDA 12.x) |
 | Pillow | 11.3.0 |
-For best reproduction we recommend running on NVIDIA H20 GPUs (matching the training/inference setup). Other GPU models may produce numerically minor (sub-percent) divergence in score outputs due to bf16 kernel-level rounding; the VQA answer is unaffected, and overall leaderboard metrics differ by less than 0.001.
 ## License

 ## Quick start
 ```bash
 pip install -r inference/requirements.txt
 # Run inference on the official Track-1 test set:
 This produces `submission.json` and `submission.zip` in the repo root.
 ## Inference
 The pipeline at inference time uses **2-pass test-time augmentation**:
 }
 ```
 ## Environment
 Pinned versions for reproducibility (see `inference/requirements.txt`):
 | torch | 2.10.0 (CUDA 12.x) |
 | Pillow | 11.3.0 |
+For best reproduction we recommend running on NVIDIA H20 GPUs (matching the training/inference setup).
 ## License