Python Execution
Script to GPU.
One command.
No Docker, no config files, no setup. Just point at your Python script and run it on any GPU.
$ lyceum python run train.py
▸ Analyzing train.py...
▸ Detected: PyTorch 2.1, transformers, wandb
▸ Estimated VRAM: 42GB → Selecting A100-80GB
▸ Installing dependencies...
Training on A100-80GB...
Epoch 1/10: loss=2.341
Epoch 2/10: loss=1.892
Epoch 3/10: loss=1.456
✓ Completed in 23m 14s • $4.82
Zero config. Here's how.
Auto dependencies
Scans your imports, installs everything automatically.
Smart GPU matching
Estimates VRAM needs, picks the right hardware.
Instant iteration
No build step. Edit your script, run again immediately.
.py
train.py
Analyzing...
Need more control?
When you need custom system packages, exact reproducibility, or non-Python runtimes, Docker execution gives you full control.
Explore Docker execution