---

name: training description: "Configure, launch, and troubleshoot Dexbotic VLA supervised fine-tuning, LoRA, and distributed training backends." metadata: disco-role: operating disable-model-invocation: true license: MIT

Dexbotic training

Use this route for experiment composition, model/data/trainer configuration, SFT, LoRA, checkpoint lineage, and DDP/DeepSpeed/FSDP backend selection. Start with training workflow, then validate the selected backend with backend resolution. Route dataset schema/conversion to data-preparation and serving/evaluation to the corresponding sibling skills.

Safe operating sequence

1. Confirm the checkpoint family, action dimension, camera count/order, state convention, dataset name, output directory, and whether the job is SFT or LoRA.
2. Run scripts/inspect_training_config.py against a copied/minimal config or JSON/YAML fragment. It checks required keys and backend compatibility without launching training.
3. Select exactly one train_backend: ddp, deepspeed, fsdp, or fsdp2. Do not infer a backend from installed packages; use the resolver contract and explicit config.
4. Match normalization stats and action transforms to the dataset. Keep norm_stats.json with the checkpoint or record its deliberate external location.
5. Use torchrun/the documented launcher only after a one-batch or one-step plan is approved. Full training, model downloads, and distributed jobs are not safe verification defaults.
6. For LoRA recipes, honor the entrypoint's backend restrictions; the documented Libero LoRA recipes use DDP and reject DeepSpeed/FSDP.

The core claims require a CUDA-capable PyTorch runtime. CPU imports can validate configuration code but cannot validate VLA training behavior.