Instructions to use microsoft/MAI-DS-R1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use microsoft/MAI-DS-R1 with Transformers:

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

pipe = pipeline("text-generation", model="microsoft/MAI-DS-R1", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("microsoft/MAI-DS-R1", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("microsoft/MAI-DS-R1", trust_remote_code=True)
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

vLLM

How to use microsoft/MAI-DS-R1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "microsoft/MAI-DS-R1"
# 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/MAI-DS-R1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/microsoft/MAI-DS-R1

SGLang

How to use microsoft/MAI-DS-R1 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/MAI-DS-R1" \
    --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/MAI-DS-R1",
		"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/MAI-DS-R1" \
        --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/MAI-DS-R1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use microsoft/MAI-DS-R1 with Docker Model Runner:
```
docker model run hf.co/microsoft/MAI-DS-R1
```

Open source the post training methodology or dataset and is V3 also in the works?

#10

by jtvino - opened Apr 24, 2025

Discussion

jtvino

Apr 24, 2025

Any plans on open sourcing the methodology for post training or releasing the dataset so it can be verified?

We finetuned the MAI-DS-R1 model on a carefully curated set of ~350K blocked topics examples using various strategies:  

Collecting and filtering query keywords 
Converting the keywords into multiple questions 
Translating the questions into various languages 
Bootstrapping answers and respective Chain of Thought (CoT) for these questions using DeepSeek R1 and internal models.

Also is MAI-DS-V3-0324 version of this in the works

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment