# BLOOMZ, a version for Petals

This model is a version of [bigscience/bloomz](https://huggingface.co/bigscience/bloomz)
post-processed to be run at home using the [Petals](https://github.com/bigscience-workshop/petals#readme) swarm.

Please check out:

- The [original model card](https://huggingface.co/bigscience/bloomz)
  to learn about the model's capabilities, specifications, and terms of use.
- The [Petals repository](https://github.com/bigscience-workshop/petals#readme)
  to learn how to install Petals and run this model over the Petals swarm.

We provide minimal code examples below.

## Using the model

```python
from petals import DistributedBloomForCausalLM

model = DistributedBloomForCausalLM.from_pretrained("bigscience/bloomz-petals")
# Embeddings & prompts are on your device, BLOOM blocks are distributed across the Internet

inputs = tokenizer("A cat sat", return_tensors="pt")["input_ids"]
outputs = model.generate(inputs, max_new_tokens=5)
print(tokenizer.decode(outputs[0]))  # A cat sat on a mat...
```

## Serving the model blocks

```bash
python -m petals.cli.run_server bigscience/bloomz-petals
```