# Local GPU Quick Start

Your **RTX 6000 Ada 48GB** is ready for Wan2.2 training!

## One-Time Setup

```bash
# 1. Install GPU environment
bash tools/setup-local-gpu-final.sh

# 2. Clone DiffSynth-Studio
git clone https://github.com/modelscope/DiffSynth-Studio.git ~/src/DiffSynth-Studio
cd ~/src/DiffSynth-Studio
pip install -e .

# 3. Download Wan2.2 model (requires HF account)
# First accept terms at: https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B
huggingface-cli login
huggingface-cli download Wan-AI/Wan2.2-TI2V-5B --local-dir ~/models/Wan2.2-TI2V-5B
```

## Run Smoke Test

```bash
# Activate GPU environment and run test
source tools/activate-gpu-env.sh
bash tools/run-smoke-test.sh

# Or run training directly:
python3 tools/run_wan22_train.py \
  --training-data-dir materials/training-data/iiw-english-smoke-video-only \
  --output-dir materials/training-data/iiw-english-smoke-video-only/wan22_checkpoints \
  --diffsynth-path $HOME/src/DiffSynth-Studio \
  --wan22-model $HOME/models/Wan2.2-TI2V-5B \
  --model-variant ti2v-5b \
  --lora-rank 16 \
  --epochs 1 \
  --dataset-repeat 20 \
  --learning-rate 2e-5 \
  --num-frames 81 \
  --height 480 \
  --width 832 \
  --gradient-accumulation-steps 4
```

**Expected time:** ~2-3 hours for 160-clip smoke test

## Monitor Training

In another terminal:
```bash
watch -n 5 nvidia-smi
```

## Checkpoints

```bash
ls -lh materials/training-data/iiw-english-smoke-video-only/wan22_checkpoints/
```

## Next Steps

After smoke test succeeds:
1. Validate output quality with test inference
2. Run full pilot training (601 clips, ~8-12 hours)
3. Generate validation shots (TSV-001, TSV-002, TSV-005)