Visual tensor graph
Connect features, reusable transforms, model blocks, targets, and losses while P³ validates tensor contracts.
04 · Build and train
ML Studio
Available in P³Design tensor flows visually, then train regression or classification models on your own hardware.
Keep the scientific context attached to the result.
P³ freezes data preparation, model structure, and training settings into reproducible versions handled by local ML Agents.
What ML Studio handles
These capabilities reflect the current product. Experimental or unsupported feature types are not presented as finished.
Regression and classification
Use one model system with task-specific heads, losses, class mappings, and evaluation metrics.
Local PyTorch Agents
Train on CPU or GPU through the managed ML runtime without handing dataset credentials to a remote service.
Version lineage
Track changes to data, architecture, and training while keeping successful trained versions immutable.
A clear sequence with inspectable handoffs.
01
Prepare an ML dataset in Analytics
02
Connect tensors and configure the model
03
Choose an Agent and start training
04
Compare metrics and retain a trained version
Start with P³
Build with P³.
Free for private and eligible university use. Companies can request a commercial license.