Instructions to use SynLayers/Bbox-caption-8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SynLayers/Bbox-caption-8b with Transformers:

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

pipe = pipeline("image-text-to-text", model="SynLayers/Bbox-caption-8b")
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, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("SynLayers/Bbox-caption-8b")
model = AutoModelForImageTextToText.from_pretrained("SynLayers/Bbox-caption-8b")
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

vLLM

How to use SynLayers/Bbox-caption-8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SynLayers/Bbox-caption-8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SynLayers/Bbox-caption-8b",
		"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/SynLayers/Bbox-caption-8b

SGLang

How to use SynLayers/Bbox-caption-8b 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 "SynLayers/Bbox-caption-8b" \
    --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": "SynLayers/Bbox-caption-8b",
		"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 "SynLayers/Bbox-caption-8b" \
        --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": "SynLayers/Bbox-caption-8b",
		"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 SynLayers/Bbox-caption-8b with Docker Model Runner:
```
docker model run hf.co/SynLayers/Bbox-caption-8b
```

SynLayers commited on 7 days ago

Commit

84ea29f

verified ·

1 Parent(s): b752efc

Upload infer/infer.yaml with huggingface_hub

Browse files

Files changed (1) hide show

infer/infer.yaml +64 -0

infer/infer.yaml ADDED Viewed

	@@ -0,0 +1,64 @@

+# Main real-world inference configuration
+seed: 42
+max_layer_num: 52
+# Size configuration
+source_size: 1024
+target_size: 1024
+# Real-world inference defaults
+data_dir: "/project/llmsvgen/share/data/kmw_layered_dataset/real_world_inference"
+image_dir: "/project/llmsvgen/share/data/kmw_layered_dataset/real_world_inference/layers_real_test_1024"
+test_jsonl: "/project/llmsvgen/share/data/kmw_layered_dataset/real_world_inference/caption_bbox_infer.jsonl"
+# Model paths
+pretrained_model_name_or_path: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_checkpoints/FLUX.1-dev"
+pretrained_adapter_path: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_checkpoints/FLUX.1-dev-Controlnet-Inpainting-Alpha"
+transp_vae_path: "ckpt/trans_vae/0008000.pt"
+# Pre-trained LoRA weights
+pretrained_lora_dir: "ckpt/pre_trained_LoRA"
+artplus_lora_dir: "ckpt/prism_ft_LoRA"
+# below is for 18k dataset
+#lora_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_18k/step_90000/transformer"
+#layer_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_18k/step_90000"
+#adapter_lora_dir: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_18k/step_90000/adapter"
+# below is for 20k dataset
+#lora_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_20k/step_120000/transformer"
+#layer_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_20k/step_120000"
+#adapter_lora_dir: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_20k/step_120000/adapter"
+# below is for 30k dataset
+#lora_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_30k/step_150000/transformer"
+#layer_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_30k/step_150000"
+#adapter_lora_dir: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_30k/step_150000/adapter"
+# below is for 40k dataset
+#lora_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_40k/step_250000/transformer"
+#layer_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_40k/step_250000"
+#adapter_lora_dir: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_40k/step_250000/adapter"
+# below is for 50k dataset
+#lora_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_50k/step_200000/transformer"
+#layer_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_50k/step_200000"
+#adapter_lora_dir: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_train_dataset/ckpt_prism_scaleup_1024_50k/step_200000/adapter"
+# unified real-world decomposition checkpoint
+lora_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_ckpt/step_120000/transformer"
+layer_ckpt: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_ckpt/step_120000"
+adapter_lora_dir: "/project/llmsvgen/share/data/kmw_layered_checkpoint/SynLayers_ckpt/step_120000/adapter"
+# Inference settings
+cfg: 4.0
+adapter_scale: 0.9
+max_sequence_length: 1024
+save_dir: "/project/llmsvgen/share/data/kmw_layered_dataset/real_world_inference/results"
+#run_name: "step_120000"  # optional manual override
+# Sample range control (1-based indexing)
+start_idx: 1
+#end_idx: 147
+#max_samples: 147