Tasks

[justinchuby]

• VNet conversion microsoft/onnxscript#1846
• List is not supported #182
• apply_weights_lazy #192
• 'list' object has no attribute 'contiguous' #93
• Unannotated input does not show up in the signature #175
• Create a minifier #162
• 'SymBool' object is not iterable #42
• Handle node call_function op when there is no _schema (for built in functions) #147
• [Feature] Handle when an op does not have signature #47
• Dynamic axis #97
• Sequence type not assigned correctly #62
• [Feature] Integrate the rewriter and optimizer #58
• Call jit.trace correctly #82
• [Feature] Make ONNXProgram callable #46
• Tuple as model output: 'tuple' object has no attribute 'metadata_props' #35
• Check adaptive avg pool decomp
• Variadic support: Constant input of tuple is unsupported #39
• FX Type promotion #118
• Iteration over 0d tensor #40
• Gemma by finding the forward function #95
• fake export
• [ONNX] dynamo: support conditional op cond  pytorch/pytorch#117655
• Handle [] inputs #108
• Graph pass for external tensors [exporter] Create a pass to turn tensors into external tensors microsoft/onnxscript#1696
• Reveal decomp information #124
• Take the dynamo strategy