- Generated by
1.9.1
|
ZigZag - Deep Learning Hardware Design Space Exploration
This repository presents the novel version of our tried-and-tested hardware Architecture-Mapping Design Space Exploration (DSE) Framework for Deep Learning (DL) accelerators. ZigZag bridges the gap between algorithmic DL decisions and their acceleration cost on specialized accelerators through a fast and accurate hardware cost estimation.
|
Classes | |
| class | OnnxTensorCategory |
| Internal representation of ONNX tensor category. More... | |
| class | OnnxTensorType |
Namespaces | |
| zigzag.parser.onnx.utils | |
Functions | |
| ModelProto | parse_onnx_model_from_path (str onnx_model_path) |
| def | add_attribute (NodeProto node, str name, Any value) |
| def | add_branch_attribute (GraphProto graph, int branch=0) |
| GraphProto | unroll_branches (GraphProto graph) |
| def | is_dynamic (ModelProto model) |
| ModelProto | parse_dynamic_onnx_model (ModelProto model) |
| Modifies the given onnx model if there's dynamic behavior in terms of an 'If' operator. More... | |
| list[int]|int | get_attribute_ints_with_name (str name, Any attrs, list[int]|int|None default=None) |
| Return the value of an attribute of given name from the given attributes If name does not exist in attrs, the default provided by the caller is used. More... | |
| def | get_onnx_tensor_type (str name, ModelProto model) |
| def | get_node_input_output_dimension_shapes (NodeProto node, ModelProto model) |
Variables | |
| logger = logging.getLogger(__name__) | |
| string | BRANCH_ATTRIBUTE = "branch" |
1.9.1