Instructions to use Soofi-Project/Soofi-S-Instruct-Preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Soofi-Project/Soofi-S-Instruct-Preview with Transformers:

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

pipe = pipeline("text-generation", model="Soofi-Project/Soofi-S-Instruct-Preview", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Soofi-Project/Soofi-S-Instruct-Preview", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("Soofi-Project/Soofi-S-Instruct-Preview", trust_remote_code=True, 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 Soofi-Project/Soofi-S-Instruct-Preview with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Soofi-Project/Soofi-S-Instruct-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": "Soofi-Project/Soofi-S-Instruct-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Soofi-Project/Soofi-S-Instruct-Preview

SGLang

How to use Soofi-Project/Soofi-S-Instruct-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 "Soofi-Project/Soofi-S-Instruct-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": "Soofi-Project/Soofi-S-Instruct-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 "Soofi-Project/Soofi-S-Instruct-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": "Soofi-Project/Soofi-S-Instruct-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Soofi-Project/Soofi-S-Instruct-Preview with Docker Model Runner:
```
docker model run hf.co/Soofi-Project/Soofi-S-Instruct-Preview
```

Compatibility with standard PEFT / QLoRA pipelines (Unsloth, TRL) for the custom Mamba-2/MoE architecture?

by Mightysweet - opened 19 days ago

Discussion

Mightysweet

19 days ago

Hey Soofi team,

Thank you for releasing the preview models! The custom hybrid Mamba-2/MoE architecture looks very promising.

Since the model relies on custom modeling code (trust_remote_code=True), I have a question regarding fine-tuning:
Is this model out-of-the-box compatible with standard PEFT / QLoRA pipelines like Unsloth or the Hugging Face TRL library? Or does the unique combination of Mamba-2 SSM layers and MoE routing require specialized training scripts and custom implementations to apply LoRA/QLoRA successfully?

If there are any recommended workflows, existing scripts, or planned support for tools like Unsloth, I would greatly appreciate any pointers.

Thanks in advance for your time and help!

Sebastian

ypipe

about 23 hours ago

Hi @Mightysweet ,

Just curious how you got access to the model. Are you under NDA or how was the process as most people seem to be blocked out?

Best
Tim

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment