Instructions to use trollek/Qwen2-0.5B-DiffusionPrompter-v0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trollek/Qwen2-0.5B-DiffusionPrompter-v0.1 with Transformers:

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

pipe = pipeline("text-generation", model="trollek/Qwen2-0.5B-DiffusionPrompter-v0.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("trollek/Qwen2-0.5B-DiffusionPrompter-v0.1")
model = AutoModelForCausalLM.from_pretrained("trollek/Qwen2-0.5B-DiffusionPrompter-v0.1", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use trollek/Qwen2-0.5B-DiffusionPrompter-v0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "trollek/Qwen2-0.5B-DiffusionPrompter-v0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "trollek/Qwen2-0.5B-DiffusionPrompter-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/trollek/Qwen2-0.5B-DiffusionPrompter-v0.1

SGLang

How to use trollek/Qwen2-0.5B-DiffusionPrompter-v0.1 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 "trollek/Qwen2-0.5B-DiffusionPrompter-v0.1" \
    --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": "trollek/Qwen2-0.5B-DiffusionPrompter-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "trollek/Qwen2-0.5B-DiffusionPrompter-v0.1" \
        --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": "trollek/Qwen2-0.5B-DiffusionPrompter-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use trollek/Qwen2-0.5B-DiffusionPrompter-v0.1 with Docker Model Runner:
```
docker model run hf.co/trollek/Qwen2-0.5B-DiffusionPrompter-v0.1
```

Qwen2-0.5B-DiffusionPrompter-v0.1 / README.md

trollek

Update README.md

f6d0415 verified about 2 years ago

preview code

Raw

History Blame Contribute Delete

1.64 kB

	---
	license: apache-2.0
	datasets:
	- Gustavosta/Stable-Diffusion-Prompts
	- trollek/Mouse-Diffusion-Instruct
	base_model: Qwen/Qwen2-0.5B
	tags:
	- llama-factory
	- unsloth
	language:
	- en
	---
	# Qwen2-0.5B-DiffusionPrompter-v0.1

	A small SD prompt model. Give it a prompt like "a beach ball" and you'll get something like:

	> A group of children play on a beach, wearing colorful clothing and hats. The sun is setting, casting a warm orange glow over the scene. The air is filled with laughter and screams as they jump and roll around in the sand. A small beach ball floats in the water, adding to the serene ambiance of the scene.

	or "painting of a beach ball, WLOP":

	> beautiful sunset on the ocean in the style of wlop, painting of a beach ball, detailed and intricate details, realistic

	## Template

	```jinja
	<\|im_start\|>user
	{{image_concept}}<\|im_end\|>
	<\|im_start\|>assistant
	{{expanded_sd_response}}<\|im_end\|>
	```

	## Quants and Ollama
	### GGUF
	- [trollek/Qwen2-0.5B-DiffusionPrompter-v0.1-GGUF](https://huggingface.co/trollek/Qwen2-0.5B-DiffusionPrompter-v0.1-GGUF)

	### Ollama

	```bash
	ollama pull trollek/qwen2-diffusion-prompter:v01-q4_K_S
	ollama pull trollek/qwen2-diffusion-prompter:v01-q5_K_S
	ollama pull trollek/qwen2-diffusion-prompter:v01-q6_K
	```

	## Notes and gratitude

	Thanks to [roborovski](https://huggingface.co/roborovski) for [superprompt-v1](https://huggingface.co/roborovski/superprompt-v1) and [Gustavosta](https://huggingface.co/Gustavosta) for [Stable-Diffusion-Prompts](https://huggingface.co/datasets/Gustavosta/Stable-Diffusion-Prompts).

	I have no idea why the parameter count has grown be 100M parameters.