Shared Infrastructure
Infrastructure shared across all ULTRATHINK research projects.
Shared Storage
/mnt/shared/
├── datasets/ # Shared datasets (ImageNet, CIFAR, etc.)
├── checkpoints/ # Model checkpoints per project
├── pretrained/ # Pretrained models (TORCH_HOME)
├── logs/ # TensorBoard, MLflow logs
└── results/ # Experiment results
Environment Variables
export {PROJECT}_DATA=/mnt/shared/datasets
export {PROJECT}_CHECKPOINTS=/mnt/shared/checkpoints/{project}
export {PROJECT}_LOGS=/mnt/shared/logs/{project}
export TORCH_HOME=/mnt/shared/pretrained
Services
| Service | Purpose | Access |
|---|---|---|
| MLflow | Experiment tracking | ai-ml namespace |
| JupyterHub | Interactive notebooks | jupyterhub namespace |
| TensorBoard | Training visualization | Docker Compose |
| Ollama | LLM inference | ai-ml namespace |
Compute Resources
| Machine | Use For |
|---|---|
| MEGATRON (2x RTX 3090) | Full training, distributed experiments |
| Lambda Laptop (1x RTX) | Ablations, single-GPU experiments |
| Mac Studio (M2 Ultra) | Inference, visualization, large model loading |
| MacBook M4 (M4 Max) | Development, quick tests |
Review Orchestrator
python /home/lucas/Papers/ultrathink_review_orchestrator.py status
python /home/lucas/Papers/ultrathink_review_orchestrator.py launch
python /home/lucas/Papers/ultrathink_review_orchestrator.py aggregate
python /home/lucas/Papers/ultrathink_review_orchestrator.py report
TODO: expand with dataset inventory, checkpoint management, and resource scheduling.