---

name: training description: "Train ALAE and StyleALAE with configs, TFRecords, DDP, and checkpoints." disable-model-invocation: true metadata: disco-role: operating license: NO_LICENSE

Training

Use this sub-skill when the request is about training or resuming ALAE/StyleALAE with train_alae.py, YACS configs, the multi-GPU launcher, TFRecords loading, LOD scheduling, checkpoint handling, or safe command construction.

Do not use this sub-skill for raw data conversion, generation/style mixing, or metrics. The README's ablation route (train_alae_separate.py, model_separate.py, and the celeba_ablation_*.yaml files) is not present in this checkout; treat it as unavailable rather than routing there.

Read first

• Training workflow
• API reference
• Checkpoint conventions
• Troubleshooting
• Config inspector

Typical outcomes

• Build a valid python train_alae.py -c <config> [YACS opts...] command.
• Check dataset, checkpoint, and output paths before starting a run.
• Diagnose GPU, DDP, TFRecords, config-key, or checkpoint mismatch issues.