Instructions to use 2stacks/s1.1-0.5B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 2stacks/s1.1-0.5B with Transformers:

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

pipe = pipeline("text-generation", model="2stacks/s1.1-0.5B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("2stacks/s1.1-0.5B")
model = AutoModelForCausalLM.from_pretrained("2stacks/s1.1-0.5B", 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 2stacks/s1.1-0.5B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "2stacks/s1.1-0.5B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "2stacks/s1.1-0.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/2stacks/s1.1-0.5B

SGLang

How to use 2stacks/s1.1-0.5B 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 "2stacks/s1.1-0.5B" \
    --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": "2stacks/s1.1-0.5B",
		"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 "2stacks/s1.1-0.5B" \
        --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": "2stacks/s1.1-0.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use 2stacks/s1.1-0.5B with Docker Model Runner:
```
docker model run hf.co/2stacks/s1.1-0.5B
```

Improve language tag

by lbourdois - opened Apr 28, 2025

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

+38

-24

Files changed (1) hide show

README.md +38 -24

README.md CHANGED Viewed

@@ -1,24 +1,38 @@
----
-pipeline_tag: text-generation
-inference: true
-license: apache-2.0
-datasets:
-- simplescaling/s1K-1.1
-base_model:
-- Qwen/Qwen2.5-0.5B-Instruct
-library_name: transformers
----
-# Model Summary
-> s1.1-0.5B is a sucessor of [s1](https://huggingface.co/2stacks/s1-0.5B) with better reasoning performance by leveraging reasoning traces from r1 instead of Gemini. This model was created simply to test the process used to train the original s1.1 cited below using consumer grade GPUs.
-- **Logs:** https://wandb.ai/2stacks-sms/s1/runs/ishervdt?nw=nwuser2stacks
-- **Repository:** [simplescaling/s1](https://github.com/simplescaling/s1)
-- **Paper:** https://arxiv.org/abs/2501.19393
-Thanks to [Ryan Marten](https://huggingface.co/ryanmarten) for helping generate r1 traces for s1K.
-# Use
-The model usage is documented [here](https://github.com/simplescaling/s1?tab=readme-ov-file#inference).

+---
+pipeline_tag: text-generation
+inference: true
+license: apache-2.0
+datasets:
+- simplescaling/s1K-1.1
+base_model:
+- Qwen/Qwen2.5-0.5B-Instruct
+library_name: transformers
+language:
+- zho
+- eng
+- fra
+- spa
+- por
+- deu
+- ita
+- rus
+- jpn
+- kor
+- vie
+- tha
+- ara
+---
+# Model Summary
+> s1.1-0.5B is a sucessor of [s1](https://huggingface.co/2stacks/s1-0.5B) with better reasoning performance by leveraging reasoning traces from r1 instead of Gemini. This model was created simply to test the process used to train the original s1.1 cited below using consumer grade GPUs.
+- **Logs:** https://wandb.ai/2stacks-sms/s1/runs/ishervdt?nw=nwuser2stacks
+- **Repository:** [simplescaling/s1](https://github.com/simplescaling/s1)
+- **Paper:** https://arxiv.org/abs/2501.19393
+Thanks to [Ryan Marten](https://huggingface.co/ryanmarten) for helping generate r1 traces for s1K.
+# Use
+The model usage is documented [here](https://github.com/simplescaling/s1?tab=readme-ov-file#inference).