Robotics
Transformers
Safetensors
English
openvla
feature-extraction
vla
image-text-to-text
multimodal
pretraining
custom_code
Instructions to use openvla/openvla-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use openvla/openvla-7b with Transformers:
# Load model directly from transformers import AutoModelForVision2Seq model = AutoModelForVision2Seq.from_pretrained("openvla/openvla-7b", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
An issue about full-fine tuning.
#3
by WenyaoZhang - opened
Thanks for your great work.
If want to use the openvla/openvla-7b as the initialized weights for full-fine tuning, what should I do.
How should I set the ”pretrained_checkpoint“, I try to set ”pretrained_checkpoint = openvla/openvla-7b", but it raise the error as following:
raise ValueError(f"Couldn't find valid HF Hub Path {hf_path = }")
ValueError: Couldn't find valid HF Hub Path hf_path = 'openvla/openvla-dev/pretrained/openvla/openvla-7b'