Instructions to use KarlP/fast-droid with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use KarlP/fast-droid with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("KarlP/fast-droid", dtype="auto") - Notebooks
- Google Colab
- Kaggle
File size: 549 Bytes
820de6e ebcc151 820de6e ebcc151 820de6e ebcc151 820de6e ebcc151 820de6e ebcc151 820de6e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 | ---
library_name: transformers
license: apache-2.0
---
# FAST DROID Expert
A [FAST tokenizer](https://www.pi.website/research/fast), trained on the [DROID dataset](https://droid-dataset.github.io/) only. For a *universal* FAST tokenizer, applicable to any robot dataset, see [here](https://huggingface.co/physical-intelligence/fast).
For details about FAST tokenizers, see our paper on efficient action tokenization for VLA training:
[**FAST: Efficient Action Tokenization for Vision-Language-Action Models**](https://arxiv.org/abs/2501.09747)
|