Instructions to use junwatu/candi-sailor2-8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use junwatu/candi-sailor2-8b with Transformers:

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

pipe = pipeline("text-generation", model="junwatu/candi-sailor2-8b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("junwatu/candi-sailor2-8b")
model = AutoModelForMultimodalLM.from_pretrained("junwatu/candi-sailor2-8b")
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 junwatu/candi-sailor2-8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "junwatu/candi-sailor2-8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "junwatu/candi-sailor2-8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/junwatu/candi-sailor2-8b

SGLang

How to use junwatu/candi-sailor2-8b 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 "junwatu/candi-sailor2-8b" \
    --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": "junwatu/candi-sailor2-8b",
		"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 "junwatu/candi-sailor2-8b" \
        --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": "junwatu/candi-sailor2-8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use junwatu/candi-sailor2-8b with Docker Model Runner:
```
docker model run hf.co/junwatu/candi-sailor2-8b
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Candi Sailor2 8B

Fine-tuned version of sail/Sailor2-8B for Indonesian cultural heritage, especially temples (candi).

Model Details

Base model: sail/Sailor2-8B (Qwen2 architecture, 8.5B parameters)
Fine-tuning method: LoRA (r=16, alpha=32) merged into base model
Training: 2 epochs, L40S GPU, ~10 minutes
Language: Indonesian, Javanese

Training Config

Parameter	Value
Learning rate	2e-4
Batch size	1 × 8 gradient accumulation
Max seq length	2048
Quantization	4-bit NF4 (training)
Optimizer	paged_adamw_8bit
Scheduler	cosine with warmup

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "junwatu/candi-sailor2-8b",
    torch_dtype="auto",
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("junwatu/candi-sailor2-8b")

messages = [{"role": "user", "content": "Apa itu Candi Borobudur?"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.7, top_p=0.9)
response = tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True)
print(response)

Example Prompts

"Apa itu Candi Borobudur?"
"Jelaskan sejarah kerajaan Majapahit"
"Apa beda candi Hindu dan Budha?"
"Ceritakan legenda Roro Jonggrang"
"Bagaimana cara memelihara warisan budaya Indonesia?"

LoRA Adapter

The LoRA adapter is also available separately: junwatu/candi-sailor2-8b-lora

License

Apache 2.0 (same as base model)

Citation

@misc{junwatu2026candi,
  title={Candi Sailor2: Fine-tuned Model for Indonesian Cultural Heritage},
  author={junwatu},
  year={2026},
  publisher={HuggingFace},
  url={https://huggingface.co/junwatu/candi-sailor2-8b}
}

@article{sailor2,
  title={Sailor2: Advancing Multilingual Large Language Models for Southeast Asian Languages},
  author={Sea AI Lab},
  journal={arXiv preprint arXiv:2502.12982},
  year={2025}
}