Instructions to use oxyapi/oxy-1-small with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use oxyapi/oxy-1-small with Transformers:

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

pipe = pipeline("text-generation", model="oxyapi/oxy-1-small")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("oxyapi/oxy-1-small")
model = AutoModelForCausalLM.from_pretrained("oxyapi/oxy-1-small")
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 oxyapi/oxy-1-small with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "oxyapi/oxy-1-small"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "oxyapi/oxy-1-small",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/oxyapi/oxy-1-small

SGLang

How to use oxyapi/oxy-1-small 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 "oxyapi/oxy-1-small" \
    --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": "oxyapi/oxy-1-small",
		"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 "oxyapi/oxy-1-small" \
        --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": "oxyapi/oxy-1-small",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use oxyapi/oxy-1-small with Docker Model Runner:
```
docker model run hf.co/oxyapi/oxy-1-small
```

micro, small, ... medium?

by lucyknada - opened Dec 8, 2024

Discussion

lucyknada

Dec 8, 2024

Hi there! amazing little model, do you have plans to scale this to 32b? thanks

TornadoAI

Oxygen org Dec 9, 2024

Hi ! yes it's planned but I'm going to need more compute resources!

lucyknada

Dec 9, 2024

that's awesome news, thanks! re:compute; are you able to secure that? also is there any plans to release the datasets? thanks!

TornadoAI

Oxygen org Dec 10, 2024

Hello, Unfortunately the datasets will remain private! We are currently looking for partners to finance the training of the medium version (probably qwen 32B or Llama 3.3 70B I'm not sure yet)

lucyknada

Dec 10, 2024

both would be interesting to see, though 70b is generally out of reach for most, gotta know who you're targeting (16-24GB average vs 48GB niche), but glad to hear there's talks to get it trained! thanks

TornadoAI

Oxygen org Dec 12, 2024

Yes I know lol I think I'm going to go on QwQ 32B sooner, But I'm still looking for partners who would like to finance the training, I'm going to wait until the Christmas holidays to take care of all that quietly outside of class!

TornadoAI changed discussion status to closed Apr 30, 2025

lucyknada

Apr 30, 2025

@TornadoAI does that mean there's no more plans?

TornadoAI

Oxygen org Apr 30, 2025

@lucyknada

The situation at Oxygen is very complicated. We have a lot of ambition but not the means. The problem is that no one is interested in our platform to use our models. Having started as a project between friends, no one is motivated either :/

The HuggingFace organization will always remain available, but I'm not sure we'll release new models.

lucyknada

Apr 30, 2025

@TornadoAI sorry to hear, would you then consider open sourcing the datasets so that could live on?

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