Instructions to use UW/OLMo2-8B-SuperBPE-t180k with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use UW/OLMo2-8B-SuperBPE-t180k with Transformers:

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

pipe = pipeline("text-generation", model="UW/OLMo2-8B-SuperBPE-t180k")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("UW/OLMo2-8B-SuperBPE-t180k")
model = AutoModelForCausalLM.from_pretrained("UW/OLMo2-8B-SuperBPE-t180k", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use UW/OLMo2-8B-SuperBPE-t180k with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "UW/OLMo2-8B-SuperBPE-t180k"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "UW/OLMo2-8B-SuperBPE-t180k",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/UW/OLMo2-8B-SuperBPE-t180k

SGLang

How to use UW/OLMo2-8B-SuperBPE-t180k 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 "UW/OLMo2-8B-SuperBPE-t180k" \
    --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": "UW/OLMo2-8B-SuperBPE-t180k",
		"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 "UW/OLMo2-8B-SuperBPE-t180k" \
        --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": "UW/OLMo2-8B-SuperBPE-t180k",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use UW/OLMo2-8B-SuperBPE-t180k with Docker Model Runner:
```
docker model run hf.co/UW/OLMo2-8B-SuperBPE-t180k
```

Training code for Tokenizer

by amazingvince - opened Apr 7, 2025

Discussion

amazingvince

Apr 7, 2025

Awesome paper and love the idea of superBPE. I would like to try training my own superBPE tokenizer and was wondering if you guys were planning on sharing the tokenizer training code? The link https://superbpe.github.io/ says coming soon. Any idea on when that might be?

alisawuffles

University of Washington org Apr 10, 2025

Thanks for following up — the tokenizer training code has been released here! https://github.com/PythonNut/superbpe

alisawuffles changed discussion status to closed Apr 17, 2025

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