Instructions to use joon-stack/univla-simpler-success50-st2-step1000 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use joon-stack/univla-simpler-success50-st2-step1000 with Transformers:
# Load model directly from transformers import OpenVLAForActionPrediction model = OpenVLAForActionPrediction.from_pretrained("joon-stack/univla-simpler-success50-st2-step1000", dtype="auto") - Notebooks
- Google Colab
- Kaggle
joon-stack/univla-simpler-success50-st2-step1000
Full merged UniVLA checkpoint for Simpler success50 fine-tuning.
- dataset:
simpler/success50 - model:
univla_stage2_lam60k_vla50k - base VLA:
hf_univla_bridge_lam_stage2_b200_tf21_vla50k_cleanenv_step050000 - base VLA HF repo:
joon-stack/univla-base-stage2-lam60k-vla50k-step050000 - step:
1000 - latent action token length:
4 - codebook size:
16 - predicted action horizon:
10
This repository contains full merged VLA weights and the matching action_decoder-1000.pt.
- Downloads last month
- 18
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support