2026/04/21 by Jean Mercat, Sedrick Keh, Kushal Arora +6 · 2 voices
Computer Science · #Baseline (sea) #Codebase #Image stitching #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Pipeline (software) #Table (database) #Topic Modeling #Training (meteorology) #Usability #cs.AI #cs.CV #cs.LG #cs.RO #cs.SE
paper · pdf · doi:10.48550/arxiv.2604.19728
openalex publication_date 2026/04/21 · arxiv published 2026/04/21 · arxiv updated 2026/04/21 · openalex created_date 2026/04/23 · openalex updated_date 2026/07/28
We present VLA Foundry, an open-source framework that unifies LLM, VLM, and VLA training in a single codebase. Most open-source VLA efforts specialize on the action training stage, often stitching together incompatible pretraining pipelines. VLA Foundry instead provides a shared training stack with end-to-end control, from language pretraining to action-expert fine-tuning. VLA Foundry supports both from-scratch training and pretrained backbones from Hugging Face. To demonstrate the utility of our framework, we train and release two types of models: the first trained fully from scratch through our LLM-->VLM-->VLA pipeline and the second built on the pretrained Qwen3-VL backbone. We evaluate closed-loop policy performance of both models on LBM Eval, an open-data, open-source simulator. We also contribute usability improvements to the simulator and the STEP analysis tools for easier public use. In the nominal evaluation setting, our fully-open from-scratch model is on par with our prior closed-source work and substituting in the Qwen3-VL backbone leads to a strong multi-task table top manipulation policy outperforming our baseline by a wide margin. The VLA Foundry codebase is available at https://github.com/TRI-ML/vlafoundry and all multi-task model weights are released on https://huggingface.co/collections/TRI-ML/vla-foundry. Additional qualitative videos are available on the project website https://tri-ml.github.io/vlafoundry.