| --- |
| license: apache-2.0 |
| tags: |
| - robotics |
| - vla |
| - vision-language-action |
| - openvla |
| - nut_thread |
| - sim |
| pipeline_tag: robotics |
| --- |
| # FIRe-VLA-OpenVLA-NutThread-Sim |
|
|
| <!-- TODO: FIRe project one-line description and paper/code links --> |
|
|
| A **OpenVLA** vision-language-action policy fine-tuned for the **Nut threading** contact-rich |
| robotic assembly task (**Sim** domain), part of the FIRe model suite. |
|
|
| - **Framework:** OpenVLA |
| - **Task:** Nut threading (`nut_thread`) |
| - **Domain:** Sim |
| - **Checkpoint:** inference-ready (20000 steps) |
|
|
| ## Usage |
|
|
| ```python |
| # Requires the OpenVLA-OFT codebase: https://github.com/moojink/openvla-oft |
| from transformers import AutoModelForVision2Seq, AutoProcessor |
| |
| repo = "bhe1004/FIRe-VLA-OpenVLA-NutThread-Sim" |
| processor = AutoProcessor.from_pretrained(repo, trust_remote_code=True) |
| model = AutoModelForVision2Seq.from_pretrained(repo, trust_remote_code=True) |
| # OFT extras bundled in the repo: action_head, proprio_projector, |
| # lora_adapter, dataset_statistics.json |
| ``` |
|
|
| ## Related |
| - Training data: [`bhe1004/FIRe-VLA-OpenVLA-NutThread-Sim`](https://huggingface.co/datasets/bhe1004/FIRe-VLA-OpenVLA-NutThread-Sim) |
|
|
