• Initial visual expansion depth of modules to keep the initial graph small and ready for selective expansion
• Show/hide non-gradient (constant) nodes
• Show module attribute names in node labels

By default Pytorch standard or inbuilt modules' internals aren't traced. Use this flag to force trace all modules (including standard Pytorch modules) up to a desired depth

Demonstrations of how TorchVista handles models that encounter runtime errors

Models with atypical input/output objects and unusual structures

• Complex input and output
• Use dict key path if tensors are passed/returned in dicts

Real-world models

Real-world CV models

Some unusual situations

Detects and compresses repeated blocks in the graph in Sequential or ModuleList containers. When consecutive modules have matching tensor dimensions and identical internal structure, they are visually compressed with a repetition count indicator.