Aline Gondim Santos | 329f862 | 2022-11-08 08:04:22 -0300 | [diff] [blame] | 1 | import onnx |
| 2 | |
Aline Gondim Santos | bd032f8 | 2022-11-25 15:39:12 -0300 | [diff] [blame] | 3 | model = onnx.load('modelSRC/mModelEncoder.onnx') |
Aline Gondim Santos | 329f862 | 2022-11-08 08:04:22 -0300 | [diff] [blame] | 4 | output =[node.name for node in model.graph.output] |
| 5 | |
| 6 | input_all = [node.name for node in model.graph.input] |
| 7 | input_initializer = [node.name for node in model.graph.initializer] |
| 8 | net_feed_input = list(set(input_all) - set(input_initializer)) |
| 9 | |
| 10 | print('Encoder Inputs: ', net_feed_input) |
| 11 | print('Encoder Outputs: ', output) |
| 12 | print("") |
| 13 | |
| 14 | onnx.checker.check_model(model) |
| 15 | graph_def = model.graph |
| 16 | |
| 17 | inputs = graph_def.input |
| 18 | for graph_input in inputs: |
| 19 | input_shape = [] |
| 20 | for d in graph_input.type.tensor_type.shape.dim: |
| 21 | if d.dim_value == 0: |
| 22 | input_shape.append(None) |
| 23 | else: |
| 24 | input_shape.append(d.dim_value) |
| 25 | print( |
| 26 | f"Input Name: {graph_input.name}, Input Data Type: {graph_input.type.tensor_type.elem_type}, Input Shape: {input_shape}" |
| 27 | ) |
| 28 | |
| 29 | |
| 30 | print("") |
| 31 | outputs = graph_def.output |
| 32 | for graph_output in outputs: |
| 33 | output_shape = [] |
| 34 | for d in graph_output.type.tensor_type.shape.dim: |
| 35 | if d.dim_value == 0: |
| 36 | output_shape.append(None) |
| 37 | else: |
| 38 | output_shape.append(d.dim_value) |
| 39 | print( |
| 40 | f"Output Name: {graph_output.name}, Output Data Type: {graph_output.type.tensor_type.elem_type}, Output Shape: {output_shape}" |
| 41 | ) |
| 42 | |
| 43 | |
Aline Gondim Santos | bd032f8 | 2022-11-25 15:39:12 -0300 | [diff] [blame] | 44 | model = onnx.load('modelSRC/mModelDecoder.onnx') |
Aline Gondim Santos | 329f862 | 2022-11-08 08:04:22 -0300 | [diff] [blame] | 45 | output =[node.name for node in model.graph.output] |
| 46 | |
| 47 | input_all = [node.name for node in model.graph.input] |
| 48 | input_initializer = [node.name for node in model.graph.initializer] |
| 49 | net_feed_input = list(set(input_all) - set(input_initializer)) |
| 50 | |
| 51 | print("\n") |
| 52 | |
| 53 | print('Decoder Inputs: ', net_feed_input) |
| 54 | print('Decoder Outputs: ', output) |
| 55 | print("") |
| 56 | |
| 57 | onnx.checker.check_model(model) |
| 58 | graph_def = model.graph |
| 59 | |
| 60 | inputs = graph_def.input |
| 61 | for graph_input in inputs: |
| 62 | input_shape = [] |
| 63 | for d in graph_input.type.tensor_type.shape.dim: |
| 64 | if d.dim_value == 0: |
| 65 | input_shape.append(None) |
| 66 | else: |
| 67 | input_shape.append(d.dim_value) |
| 68 | print( |
| 69 | f"Input Name: {graph_input.name}, Input Data Type: {graph_input.type.tensor_type.elem_type}, Input Shape: {input_shape}" |
| 70 | ) |
| 71 | |
| 72 | print("") |
| 73 | outputs = graph_def.output |
| 74 | for graph_output in outputs: |
| 75 | output_shape = [] |
| 76 | for d in graph_output.type.tensor_type.shape.dim: |
| 77 | if d.dim_value == 0: |
| 78 | output_shape.append(None) |
| 79 | else: |
| 80 | output_shape.append(d.dim_value) |
| 81 | print( |
| 82 | f"Output Name: {graph_output.name}, Output Data Type: {graph_output.type.tensor_type.elem_type}, Output Shape: {output_shape}" |
| 83 | ) |