Edward56 LingBot-VLA Workspace for OpenRoboto SN80
This repository is maintained by edward56 as a complete, evaluator-ready
LingBot-VLA checkpoint workspace for the OpenRoboto simulation track on
Bittensor Subnet 80.
The model is published as a fully merged checkpoint. It is not a standalone adapter. The model shards, index, processor files, tokenizer files, and runtime configuration remain together at the repository root so the evaluator can load the checkpoint directly.
Repository Layout
docs/contains the maintainer handoff and usage notes.evaluation/preserves evaluation material distributed with the artifact.provenance/preserves traceability records for the checkpoint package.training/preserves the training-side configuration and normalization data needed to understand and reproduce the artifact.
Compatibility Policy
Runtime-sensitive files are intentionally kept unchanged unless a genuinely new trained checkpoint is produced. In particular, camera mapping, action semantics, normalization statistics, model configuration, and tokenizer files must stay mutually compatible.
Maintainer Note
The documentation in this repository is maintained in English by edward56.
The preserved manifests and provenance files describe the artifact history;
this model card does not replace or rewrite those records.
For the current submission rules and checkpoint checks, use the OpenRoboto documentation and the LingBot miner guide.
- Downloads last month
- -