Instructions to use Tiiny/SmallThinker-3B-Preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Tiiny/SmallThinker-3B-Preview with Transformers:

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

pipe = pipeline("text-generation", model="Tiiny/SmallThinker-3B-Preview")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Tiiny/SmallThinker-3B-Preview")
model = AutoModelForCausalLM.from_pretrained("Tiiny/SmallThinker-3B-Preview")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Tiiny/SmallThinker-3B-Preview with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Tiiny/SmallThinker-3B-Preview"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Tiiny/SmallThinker-3B-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Tiiny/SmallThinker-3B-Preview

SGLang

How to use Tiiny/SmallThinker-3B-Preview 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 "Tiiny/SmallThinker-3B-Preview" \
    --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": "Tiiny/SmallThinker-3B-Preview",
		"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 "Tiiny/SmallThinker-3B-Preview" \
        --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": "Tiiny/SmallThinker-3B-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Tiiny/SmallThinker-3B-Preview with Docker Model Runner:
```
docker model run hf.co/Tiiny/SmallThinker-3B-Preview
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

SmallThinker-3B-preview

We introduce SmallThinker-3B-preview, a new model fine-tuned from the Qwen2.5-3b-Instruct model.

Now you can directly deploy SmallThinker On your phones with PowerServe.

Benchmark Performance

Model	AIME24	AMC23	GAOKAO2024_I	GAOKAO2024_II	MMLU_STEM	AMPS_Hard	math_comp
Qwen2.5-3B-Instruct	6.67	45	50	35.8	59.8	-	-
SmallThinker	16.667	57.5	64.2	57.1	68.2	70	46.8
GPT-4o	9.3	-	-	-	64.2	57	50

Limitation: Due to SmallThinker's current limitations in instruction following, for math_comp we adopt a more lenient evaluation method where only correct answers are required, without constraining responses to follow the specified AAAAA format.

Colab Link: Colab

Intended Use Cases

SmallThinker is designed for the following use cases:

Edge Deployment: Its small size makes it ideal for deployment on resource-constrained devices.
Draft Model for QwQ-32B-Preview: SmallThinker can serve as a fast and efficient draft model for the larger QwQ-32B-Preview model. From my test, in llama.cpp we can get 70% speedup (from 40 tokens/s to 70 tokens/s).

Training Details

The model was trained using 8 H100 GPUs with a global batch size of 16. The specific configuration is as follows:

The SFT (Supervised Fine-Tuning) process was conducted in two phases:

First Phase:
- Used only the PowerInfer/QWQ-LONGCOT-500K dataset
- Trained for 1.5 epochs

### model
model_name_or_path: /home/syx/Qwen2.5-3B-Instruct

### method
stage: sft
do_train: true
finetuning_type: full
deepspeed: examples/deepspeed/ds_z3_config.json

### dataset
dataset: o1-v2
template: qwen
neat_packing: true
cutoff_len: 16384
overwrite_cache: true
preprocessing_num_workers: 16

### output
output_dir: saves/qwen2-01-qat/full/sft
logging_steps: 1
save_steps: 1000
plot_loss: true
overwrite_output_dir: true

Second Phase:
- Combined training with PowerInfer/QWQ-LONGCOT-500K and PowerInfer/LONGCOT-Refine datasets
- Continued training for 2 additional epochs

### model
model_name_or_path: saves/qwen2-01-qat/full/sft/checkpoint-24000

### method
stage: sft
do_train: true
finetuning_type: full
deepspeed: examples/deepspeed/ds_z3_config.json

### dataset
dataset: o1-v2, o1-v3
template: qwen
neat_packing: true
cutoff_len: 16384
overwrite_cache: true
preprocessing_num_workers: 16

### output
output_dir: saves/qwen2-01-qat/full/sft
logging_steps: 1
save_steps: 1000
plot_loss: true
overwrite_output_dir: true

Limitations & Disclaimer

Please be aware of the following limitations:

Language Limitation: The model has only been trained on English-language datasets, hence its capabilities in other languages are still lacking.
Limited Knowledge: Due to limited SFT data and the model's relatively small scale, its reasoning capabilities are constrained by its knowledge base.
Unpredictable Outputs: The model may produce unexpected outputs due to its size and probabilistic generation paradigm. Users should exercise caution and validate the model's responses.
Repetition Issue: The model tends to repeat itself when answering high-difficulty questions. Please increase the repetition_penalty to mitigate this issue.