---
tags:
  - bart
  - mirror
library_name: transformers
license: apache-2.0
---

# BART-large (full mirror)

Full mirror of [facebook/bart-large](https://huggingface.co/facebook/bart-large) — includes model weights (`pytorch_model.bin`, ~971 MB), tokenizer files, config.

Mirrored via `huggingface_hub.snapshot_download` for archival; identical to the upstream snapshot at mirror time.

## Usage

```python
from transformers import AutoModel, AutoTokenizer
model = AutoModel.from_pretrained("arrow-hf/bart-large")
tokenizer = AutoTokenizer.from_pretrained("arrow-hf/bart-large")
```

## Related

The tokenizer is used by [arrow-hf/xvla-robotwin-stack-bowls-two-40pct](https://huggingface.co/arrow-hf/xvla-robotwin-stack-bowls-two-40pct) (max_length=50). The X-VLA model uses the BART tokenizer vocabulary; the Florence2 vision-language backbone has its own learned weights and does not load BART's weights directly.