Instructions to use microsoft/UserRepGen-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use microsoft/UserRepGen-4B with Transformers:

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

pipe = pipeline("text-generation", model="microsoft/UserRepGen-4B")
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("microsoft/UserRepGen-4B")
model = AutoModelForMultimodalLM.from_pretrained("microsoft/UserRepGen-4B", 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 microsoft/UserRepGen-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "microsoft/UserRepGen-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "microsoft/UserRepGen-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/microsoft/UserRepGen-4B

SGLang

How to use microsoft/UserRepGen-4B 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 "microsoft/UserRepGen-4B" \
    --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": "microsoft/UserRepGen-4B",
		"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 "microsoft/UserRepGen-4B" \
        --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": "microsoft/UserRepGen-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use microsoft/UserRepGen-4B with Docker Model Runner:
```
docker model run hf.co/microsoft/UserRepGen-4B
```

UserRepGen-4B / representation_prompt.txt

philippelaban

Copilot

Add teaser, inference assets, and working UserRepGen example

4b75a7e verified 4 days ago

Raw History Blame Contribute Delete

1.47 kB

	You are an expert at analyzing user behavior in conversations between a user and an AI assistant. Given the conversation below, your task is to produce a representation of the user.

	Treat the conversation as evidence about the user, and use it to infer how they generally behave and how they would typically act in other conversations. Focus on stable, general tendencies such as: how they phrase and specify requests, how they decompose goals across turns, how they respond to and give feedback to the assistant, how and when they provide context, their communication style, the dialog acts they use, and their degree and depth of engagement. Keep this description general: do not include topic-specific details or content from the conversation, do not quote or closely paraphrase full user messages, and do not refer to specific moments in this conversation (e.g. "here", "in this conversation", "when they asked about X"). Short distinctive words or phrases may be quoted sparingly when they are useful examples of the user's communication style. Describe their patterns in your own general terms instead.

	Separately, you will also describe the task the user is trying to accomplish. Here you may include topic-specific details about what the user is doing in the conversation.

	Conversation:
	{conversation_history}

	Output a single valid JSON object of the form {{"task": "...", "user_behavior": "..."}}. Do not wrap the output in markdown or add any other text.