@reicek/neataptic-ts 0.1.24 → 0.1.26

package/src/architecture/network/onnx/schema/README.md

@@ -38,13 +38,13 @@ Pitfall: mappings must match the actual layer sizes. If `inHeight * inWidth * in
 does not correspond to the prior layer width (and similarly for outputs), export or import
 may reject the model.
 
-### Pool2DMapping
+### OnnxAttribute
 
-Mapping describing a pooling operation inserted after a given export-layer index.
+ONNX node attribute payload.
 
-This is represented as metadata and optional graph nodes during export.
-Import uses it to attach pooling-related runtime metadata back onto the reconstructed
-network (when supported).
+This simplified JSON-first shape is enough for the operators emitted by the
+current exporter. It intentionally avoids protobuf-level complexity while
+still preserving the attribute variants needed by the importer.
 
 ### OnnxDimension
 
@@ -53,25 +53,18 @@ One dimension inside an ONNX tensor shape.
 Use `dim_value` for fixed numeric widths and `dim_param` for symbolic names
 such as a batch dimension.
 
-### OnnxShape
-
-ONNX tensor type shape.
-
-### OnnxTensorType
-
-ONNX tensor type.
-
-### OnnxValueInfo
+### OnnxGraph
 
-ONNX value info (input/output description).
+Graph body of an ONNX-like model.
 
-### OnnxAttribute
+The exporter writes three main collections here:
+- `inputs` and `outputs` describe graph boundaries,
+- `initializer` stores constant tensors such as weights and biases,
+- `node` stores the ordered operator payloads that consume those tensors.
 
-ONNX node attribute payload.
+### OnnxMetadataProperty
 
-This simplified JSON-first shape is enough for the operators emitted by the
-current exporter. It intentionally avoids protobuf-level complexity while
-still preserving the attribute variants needed by the importer.
+Canonical metadata key-value pair used in ONNX model metadata_props.
 
 ### OnnxModel
 
@@ -92,14 +85,16 @@ Notes:
 Security/trust boundary:
 - Treat this as untrusted input if it comes from outside your process.
 
-### OnnxGraph
+### OnnxNode
 
-Graph body of an ONNX-like model.
+One ONNX operator invocation inside the graph.
 
-The exporter writes three main collections here:
-- `inputs` and `outputs` describe graph boundaries,
-- `initializer` stores constant tensors such as weights and biases,
-- `node` stores the ordered operator payloads that consume those tensors.
+Nodes connect named tensors rather than object references, which keeps the
+exported payload easy to serialize, inspect, and diff as plain JSON.
+
+### OnnxShape
+
+ONNX tensor type shape.
 
 ### OnnxTensor
 
@@ -108,13 +103,18 @@ Serialized tensor payload stored inside graph initializers.
 NeatapticTS currently writes floating-point parameter vectors and matrices to
 `float_data`, along with the tensor name, element type, and logical shape.
 
-### OnnxNode
+### OnnxTensorType
 
-One ONNX operator invocation inside the graph.
+ONNX tensor type.
 
-Nodes connect named tensors rather than object references, which keeps the
-exported payload easy to serialize, inspect, and diff as plain JSON.
+### OnnxValueInfo
 
-### OnnxMetadataProperty
+ONNX value info (input/output description).
 
-Canonical metadata key-value pair used in ONNX model metadata_props.
+### Pool2DMapping
+
+Mapping describing a pooling operation inserted after a given export-layer index.
+
+This is represented as metadata and optional graph nodes during export.
+Import uses it to attach pooling-related runtime metadata back onto the reconstructed
+network (when supported).