Instructions to use DeepBrainz/DeepBrainz-R1-4B-16K with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DeepBrainz/DeepBrainz-R1-4B-16K with Transformers:

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

pipe = pipeline("text-generation", model="DeepBrainz/DeepBrainz-R1-4B-16K")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DeepBrainz/DeepBrainz-R1-4B-16K")
model = AutoModelForCausalLM.from_pretrained("DeepBrainz/DeepBrainz-R1-4B-16K", 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 DeepBrainz/DeepBrainz-R1-4B-16K with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DeepBrainz/DeepBrainz-R1-4B-16K"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DeepBrainz/DeepBrainz-R1-4B-16K",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/DeepBrainz/DeepBrainz-R1-4B-16K

SGLang

How to use DeepBrainz/DeepBrainz-R1-4B-16K 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 "DeepBrainz/DeepBrainz-R1-4B-16K" \
    --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": "DeepBrainz/DeepBrainz-R1-4B-16K",
		"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 "DeepBrainz/DeepBrainz-R1-4B-16K" \
        --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": "DeepBrainz/DeepBrainz-R1-4B-16K",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DeepBrainz/DeepBrainz-R1-4B-16K with Docker Model Runner:
```
docker model run hf.co/DeepBrainz/DeepBrainz-R1-4B-16K
```

ArunkumarVR commited on Jan 30

Commit

01981d4

verified ·

1 Parent(s): b02b2e2

Update README.md

Browse files

Files changed (1) hide show

README.md +4 -3

README.md CHANGED Viewed

@@ -26,13 +26,14 @@ The model emphasizes **reasoning quality, instruction robustness, and stability
 - 16K context length
 - Optimized for reasoning-centric tasks
 - Designed for modern GPU inference runtimes
-- **Architecture:** Qwen3-compatible (DeepBrainz-R series, optimized via OPD)
 ---
 ## Intended Use
-- Advanced reasoning systems
 - Research and evaluation
 - Agentic workflows
 - Inference-time scaling and test-time compute experiments
@@ -69,7 +70,7 @@ print(tok.decode(out[0], skip_special_tokens=True))
 ## Training Summary
-The model was produced using a **multi-stage optimization process** involving large-scale supervision and **iterative refinement** to improve reasoning quality and robustness.
 Specific training details are intentionally abstracted in this public release.

 - 16K context length
 - Optimized for reasoning-centric tasks
 - Designed for modern GPU inference runtimes
+- **Architecture:** Qwen3-compatible (DeepBrainz-R series, post-trained, and optimized for math and coding)
 ---
 ## Intended Use
+- Advanced reasoning systems
+- Math and Coding
 - Research and evaluation
 - Agentic workflows
 - Inference-time scaling and test-time compute experiments
 ## Training Summary
+The model was produced using a **multi-stage optimization process** involving large-scale on-policy optimization and **iterative refinement** to improve reasoning quality and robustness.
 Specific training details are intentionally abstracted in this public release.