NVIDIA FastGen training
Opens the skill on ClawHub, which shows its security scan.
Source ClawHub, published by nvidia
Overview
This NVIDIA skill helps an OpenClaw agent run FastGen training: supervised fine-tuning or distillation of diffusion models with the project's train.py. It covers short smoke tests, distributed launches across GPUs, checkpoints and resuming, validation and W&B logs as evidence. Use it when a training run needs to start, continue or be fixed after it fails. Training jobs use real GPU time and storage, so confirm the size of a run first.
Use cases
- Model training
- SFT training
- Diffusion model distillation
- Distributed training
- Model fine-tuning
Show all 7Show fewer
- PyTorch
- Weights & Biases
