Instructions to use Hyggshi-AI/HOS-OSS-3.08B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Hyggshi-AI/HOS-OSS-3.08B with Transformers:

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

pipe = pipeline("text-generation", model="Hyggshi-AI/HOS-OSS-3.08B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Hyggshi-AI/HOS-OSS-3.08B")
model = AutoModelForCausalLM.from_pretrained("Hyggshi-AI/HOS-OSS-3.08B", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Hyggshi-AI/HOS-OSS-3.08B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Hyggshi-AI/HOS-OSS-3.08B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Hyggshi-AI/HOS-OSS-3.08B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Hyggshi-AI/HOS-OSS-3.08B

SGLang

How to use Hyggshi-AI/HOS-OSS-3.08B 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 "Hyggshi-AI/HOS-OSS-3.08B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Hyggshi-AI/HOS-OSS-3.08B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "Hyggshi-AI/HOS-OSS-3.08B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Hyggshi-AI/HOS-OSS-3.08B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Hyggshi-AI/HOS-OSS-3.08B with Docker Model Runner:
```
docker model run hf.co/Hyggshi-AI/HOS-OSS-3.08B
```

HOS-OSS-3.08B / README.md

hydffgg

Update README.md

1754de2 verified 22 days ago

preview code

Raw

History Blame Contribute Delete

1.65 kB


	---
	language:
	- en
	license: apache-2.0
	library_name: transformers
	pipeline_tag: text-generation
	tags:
	- llama
	- causal-lm
	- code-generation
	- lightweight
	- 3.08B
	base_model:
	- Qwen/Qwen2.5-Coder-3B-Instruct
	---

	<p align="center">
	<img alt="HOS-OSS-3.08B" src="https://huggingface.co/hydffgg/HOS-OSS-1.54B/resolve/main/HOS-OSS-270M.png">
	</p>


	# HOS-OSS-3.08B

	HOS-OSS-3.08B is a lightweight 3.08B parameter causal language model optimized for text and code generation tasks.
	It is designed for fast inference, low resource usage, and local deployment.

	---

	## 🚀 Overview

	- Model size: ~3.08B parameters
	- Architecture: LLaMA-style decoder-only transformer
	- Base model: Qwen2.5-Coder-3B-Instruct (distilled / adapted)
	- Framework: 🤗 Transformers
	- Use cases:
	- Code generation
	- Instruction following
	- Chat-style completion
	- Lightweight local AI assistant

	---

	## ⚡ Features

	- Fast inference on low-end GPUs
	- Runs on Kaggle / Colab without large VRAM
	- Suitable for edge deployment
	- Clean instruction-response formatting

	---

	## 🧠 Example Usage

	```python
	from transformers import AutoTokenizer, AutoModelForCausalLM
	import torch

	model_name = "Hyggshi-AI/HOS-OSS-3.08B"

	tokenizer = AutoTokenizer.from_pretrained(model_name)
	model = AutoModelForCausalLM.from_pretrained(model_name)

	prompt = "User: Write a Python Hello World
	Assistant:"

	inputs = tokenizer(prompt, return_tensors="pt")

	with torch.no_grad():
	outputs = model.generate(
	**inputs,
	max_new_tokens=512,
	temperature=0.7
	)

	print(tokenizer.decode(outputs[0], skip_special_tokens=True))