Instructions to use SakanaAI/TinySwallow-1.5B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SakanaAI/TinySwallow-1.5B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="SakanaAI/TinySwallow-1.5B-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SakanaAI/TinySwallow-1.5B-Instruct")
model = AutoModelForCausalLM.from_pretrained("SakanaAI/TinySwallow-1.5B-Instruct")
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 SakanaAI/TinySwallow-1.5B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SakanaAI/TinySwallow-1.5B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SakanaAI/TinySwallow-1.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SakanaAI/TinySwallow-1.5B-Instruct

SGLang

How to use SakanaAI/TinySwallow-1.5B-Instruct 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 "SakanaAI/TinySwallow-1.5B-Instruct" \
    --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": "SakanaAI/TinySwallow-1.5B-Instruct",
		"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 "SakanaAI/TinySwallow-1.5B-Instruct" \
        --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": "SakanaAI/TinySwallow-1.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SakanaAI/TinySwallow-1.5B-Instruct with Docker Model Runner:
```
docker model run hf.co/SakanaAI/TinySwallow-1.5B-Instruct
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

TinySwallow-1.5B-Instruct

🤗 Models | 📚 Paper | 📝 Blog | 🐦 Twitter

TinySwallow-1.5B-Instruct is an instruction-tuned version of TinySwallow-1.5B, created through TAID (Temporally Adaptive Interpolated Distillation), our new knowledge distillation method. We used Qwen2.5-32B-Instruct as the teacher model and Qwen2.5-1.5B-Instruct as the student model. The model has been further instruction-tuned to enhance its ability to follow instructions and engage in conversations in Japanese.

Usage

Use the code below to get started with the model.

Click to expand

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer


# 1. load model
device = "cuda" if torch.cuda.is_available() else "cpu"
repo_id = "SakanaAI/TinySwallow-1.5B-Instruct"
model = AutoModelForCausalLM.from_pretrained(repo_id)
tokenizer = AutoTokenizer.from_pretrained(repo_id)
model.to(device)

# 2. prepare inputs
text = "知識蒸留について簡単に教えてください。"
messages = [{"role": "user", "content": text}]
input_ids = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")

# 3. generate
output_ids = model.generate(
    input_ids.to(device),
    max_new_tokens=1024,
)
output_ids = output_ids[:, input_ids.shape[1] :]
generated_text = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0]
print(generated_text)

Model Details

Model type: Autoregressive Language Model
Language(s): Japanese
Paper: https://arxiv.org/abs/2501.16937
Blog: https://sakana.ai/taid-jp/
Training Datasets:

Uses

This model is provided for research and development purposes only and should be considered as an experimental prototype. It is not intended for commercial use or deployment in mission-critical environments. Use of this model is at the user's own risk, and its performance and outcomes are not guaranteed. Sakana AI shall not be liable for any direct, indirect, special, incidental, or consequential damages, or any loss arising from the use of this model, regardless of the results obtained. Users must fully understand the risks associated with the use of this model and use it at their own discretion.

Acknowledgement

We would like to thank the developers of the source models for their contributions and for making their work available.

Authors

License

This model is derived from Qwen (Apache 2.0) and trained on Gemma data (Gemma Terms, Prohibited Use). Use (including commercial) is permitted if you comply with both licenses/policies above.

Citation

@misc{sakana2025taid,
      title         = {TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models}, 
      author.       = {Makoto Shing and Kou Misaki and Han Bao and Sho Yokoi and Takuya Akiba},
      year          = {2025},
      eprint        = {2501.16937},
      archivePrefix = {arXiv},
      primaryClass  = {cs.LG},
      url           = {https://arxiv.org/abs/2501.16937}
}