OpenVLA-OFT
OpenVLA collaboration
An optimized OpenVLA fine-tuning recipe with continuous action chunks, multi-image input, and faster high-frequency control.
- Code
- Available
- Weights
- Not listed
- Checkpoint
- Not listed
- License
- MIT
Decision summary
Fine-tuning speed
Hardware, latency, dependencies, and checkpoint loading are not pipeline-tested.
Verify weight files, sizes, format, and loading instructions.
Evidence profile
A visual read of adoption evidence. Scores describe catalog evidence readiness, not task performance.
Published evaluation
3/6
dimensions scored
Unknown evidence remains visible and is never treated as a zero.
Why these scores
The strongest decision reasons behind the evidence profile.
Artifact availability
UsefulSource files were verified, but public weights are not available.
Training and loading reproducibility
UnknownNo verified loading configuration is available.
Evaluation evidence
UnknownNo structured evaluation evidence has been verified.
Data requirements
Declared loop-data needs, missing evidence, and the next checks that matter.
Observation / ego video
observation · camera pose
Language intent / task phase
language intent
Action / robot state
actions · robot state
Sim-real / embodiment metadata
robot state · camera pose
Critical gaps
Core categories are represented. Interface alignment and data quality still require verification.
Linked datasets
Signal-level links only. Verify runtime interfaces before use.
More evidenceShowHide
Additional decision dimensions
Access and license are declared by the source.
75 · c85Required signal categories are structured; exact tensor and action interfaces still need verification.
76 · c70Hardware, latency, dependencies, and checkpoint loading are not pipeline-tested.
— · c10Artifact facts
- release_year
- official_claim
- 2025
- curated official source year
- model.required_signals
- official_claim
- observation · language intent · actions · robot state · camera pose
- Official model documentation and curated signal mapping
- repository.code
- metadata_verified
- repository: moojink/openvla-oft · default branch: main · archived: false · code file count: 101 · sample files: experiments/robot/aloha/aloha_utils.py · experiments/robot/aloha/constants.py · experiments/robot/aloha/preprocess_split_aloha_data.py · experiments/robot/aloha/real_env.py · experiments/robot/aloha/robot_utils.py · experiments/robot/aloha/run_aloha_eval.py · experiments/robot/libero/libero_utils.py · experiments/robot/libero/regenerate_libero_dataset.py · experiments/robot/libero/run_libero_eval.py · experiments/robot/openvla_utils.py · experiments/robot/robot_utils.py · prismatic/__init__.py · prismatic/conf/__init__.py · prismatic/conf/datasets.py · prismatic/conf/models.py · prismatic/conf/vla.py · prismatic/extern/__init__.py · prismatic/extern/hf/__init__.py · prismatic/extern/hf/configuration_prismatic.py · prismatic/extern/hf/modeling_prismatic.py
- GitHub repository metadata and recursive tree
- license.code
- metadata_verified
- MIT
- GitHub repository SPDX license metadata
- license
- metadata_verified
- MIT
- GitHub repository SPDX license metadata
- release_timing
- metadata_verified
- 2025-02-28T00:57:03.000Z
- GitHub repository metadata created_at
Additional references
OpenBot notes
- Treat as an OpenVLA release/recipe in the future series schema, despite the current flat static index.
