yobx.torch.interpreter

class yobx.torch.interpreter.TorchOpOverload

The class is unused only to bypass a documentation warning. The alias TorchOpOverload refers to torch._ops.Overload.

yobx.torch.interpreter.make_undefined_dimension(i: int) → torch.SymInt

Uses for a custom op when a new dimension must be introduced to bypass some verification. The following function creates a dummy output with a dimension based on the content.

def symbolic_shape(x, y):
    return torch.empty(
        x.shape[0],
        make_undefined_dimension(min(x.shape[1], y[0])),
    )

modules

• yobx.torch.interpreter._aten_functions
• yobx.torch.interpreter._aten_functions_attention
• yobx.torch.interpreter._aten_functions_transformers
• yobx.torch.interpreter._aten_methods
• yobx.torch.interpreter._doc_
• yobx.torch.interpreter._exceptions
• yobx.torch.interpreter._math_functions
• yobx.torch.interpreter._non_aten_functions
• yobx.torch.interpreter._prims_functions
• yobx.torch.interpreter._torch_helper
• yobx.torch.interpreter.aten_functions
• yobx.torch.interpreter.aten_functions_transformers
• yobx.torch.interpreter.aten_methods
• yobx.torch.interpreter.dispatcher
• yobx.torch.interpreter.interpreter
• yobx.torch.interpreter.onnx_export