214336 - rpi5 aidge – dAIEdge-VLab Status

dAIEdge-VLab Weekly Tests

rpi5

rpi5-0002-0000

aidge

214336 - rpi5 aidge

❗

Summary: Test failed.

Model Details

model name : Resnet.onnx

Logs Details

user.log

MODEL : Resnet.onnx

===============
ONNX Graph
===============

input_1 ['unk__126', 32, 32, 3]

===============
Aidge Graph
===============

  Node(name='const_fold_opt__107', optype='Producer', children: [[1]])
  Node(name='model_activation_1_Relu;model_batch_normalization_1_FusedBatchNormV3;model_conv2d_1_BiasAdd_ReadVariableOp_resource;model_conv2d_1_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_1_Conv2D', optype='Producer', children: [[1]])
  Node(name='model_activation_2_Relu;model_add_add', optype='Add', parents: [1, 1], children: [[1]])
  Node(name='Relu__12', optype='ReLU', parents: [1], children: [[1, 1]])
  Node(name='const_fold_opt__125', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__113', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__121', optype='Producer', children: [[1]])
  Node(name='model_activation_4_Relu;model_add_1_add', optype='Add', parents: [1, 1], children: [[1]])
  Node(name='Identity', optype='Softmax', parents: [1], children: [[]])
  Node(name='model_flatten_Reshape', optype='Reshape', parents: [1, 1], children: [[1]])
  Node(name='model_dense_MatMul;model_dense_BiasAdd', optype='MatMul', parents: [1, 1], children: [[1]])
  Node(name='Add__29', optype='Add', parents: [1, 1], children: [[1]])
  Node(name='model_activation_1_Relu;model_batch_normalization_1_FusedBatchNormV3;model_conv2d_1_BiasAdd_ReadVariableOp_resource;model_conv2d_1_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_1_Conv2D1', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='model_batch_normalization_4_FusedBatchNormV3;model_conv2d_4_BiasAdd_ReadVariableOp_resource;model_conv2d_4_BiasAdd', optype='Producer', children: [[1]])
  Node(name='model_activation_3_Relu;model_batch_normalization_3_FusedBatchNormV3;model_conv2d_3_BiasAdd_ReadVariableOp_resource;model_conv2d_3_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_3_Conv2D1', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='const_fold_opt__111', optype='Producer', children: [[1]])
  Node(name='model_activation_5_Relu;model_batch_normalization_5_FusedBatchNormV3;model_conv2d_6_BiasAdd_ReadVariableOp_resource;model_conv2d_6_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_6_Conv2D', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__123', optype='Producer', children: [[1]])
  Node(name='Relu__8', optype='ReLU', parents: [1], children: [[1]])
  Node(name='model_activation_Relu;model_batch_normalization_FusedBatchNormV3;model_conv2d_BiasAdd_ReadVariableOp_resource;model_conv2d_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_Conv2D', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__124', optype='Producer', children: [[1]])
  Node(name='model_batch_normalization_6_FusedBatchNormV3;model_conv2d_7_BiasAdd_ReadVariableOp_resource;model_conv2d_7_BiasAdd', optype='Producer', children: [[1]])
  Node(name='model_conv2d_5_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_5_BiasAdd_ReadVariableOp_resource1', optype='Conv2D', parents: [1, 1, 1], children: [[1]])
  Node(name='model_activation_6_Relu;model_add_2_add', optype='Add', parents: [1, 1], children: [[1]])
  Node(name='const_fold_opt__116', optype='Producer', children: [[1]])
  Node(name='model_average_pooling2d_AvgPool', optype='AvgPooling2D', parents: [1], children: [[1]])
  Node(name='model_activation_Relu;model_batch_normalization_FusedBatchNormV3;model_conv2d_BiasAdd_ReadVariableOp_resource;model_conv2d_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_Conv2D1', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='model_conv2d_8_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_8_BiasAdd_ReadVariableOp_resource', optype='Producer', children: [[1]])
  Node(name='model_activation_5_Relu;model_batch_normalization_5_FusedBatchNormV3;model_conv2d_6_BiasAdd_ReadVariableOp_resource;model_conv2d_6_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_6_Conv2D1', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='Relu__14', optype='ReLU', parents: [1], children: [[1]])
  Node(name='Relu__19', optype='ReLU', parents: [1], children: [[1, 1]])
  Node(name='model_activation_Relu;model_batch_normalization_FusedBatchNormV3;model_conv2d_BiasAdd_ReadVariableOp_resource;model_conv2d_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_Conv2D1__31', optype='Transpose', parents: [0], children: [[1]])
  Node(name='const_fold_opt__117', optype='Producer', children: [[1]])
  Node(name='model_conv2d_8_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_8_BiasAdd_ReadVariableOp_resource1', optype='Conv2D', parents: [1, 1, 1], children: [[1]])
  Node(name='Relu__5', optype='ReLU', parents: [1], children: [[1, 1]])
  Node(name='const_fold_opt__108', optype='Producer', children: [[1]])
  Node(name='model_dense_BiasAdd_ReadVariableOp_resource', optype='Producer', children: [[1]])
  Node(name='Relu__21', optype='ReLU', parents: [1], children: [[1]])
  Node(name='model_activation_3_Relu;model_batch_normalization_3_FusedBatchNormV3;model_conv2d_3_BiasAdd_ReadVariableOp_resource;model_conv2d_3_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_3_Conv2D', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__120', optype='Producer', children: [[1]])
  Node(name='Relu__26', optype='ReLU', parents: [1], children: [[1]])
  Node(name='model_conv2d_5_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_5_BiasAdd_ReadVariableOp_resource', optype='Producer', children: [[1]])
  Node(name='model_batch_normalization_6_FusedBatchNormV3;model_conv2d_7_BiasAdd_ReadVariableOp_resource;model_conv2d_7_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_7_Conv2D', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='model_batch_normalization_2_FusedBatchNormV3;model_conv2d_2_BiasAdd_ReadVariableOp_resource;model_conv2d_2_BiasAdd;model_conv2d_2_Conv2D', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='model_batch_normalization_2_FusedBatchNormV3;model_conv2d_2_BiasAdd_ReadVariableOp_resource;model_conv2d_2_BiasAdd', optype='Producer', children: [[1]])
  Node(name='model_batch_normalization_4_FusedBatchNormV3;model_conv2d_4_BiasAdd_ReadVariableOp_resource;model_conv2d_4_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_4_Conv2D', optype='PaddedConv2D', parents: [1, 1, 1, 0, 0, 0], children: [[1]])

===============
Supported nodes
===============

Native operators: 46 (10 types)
- Add: 4
- AvgPooling2D: 1
- Conv2D: 2
- MatMul: 1
- PaddedConv2D: 7
- Producer: 21
- ReLU: 7
- Reshape: 1
- Softmax: 1
- Transpose: 1
Generic operators: 0 (0 types)
Native types coverage: 100.0% (10/10)
Native operators coverage: 100.0% (46/46)
  (defaultdict(<class 'int'>, {'Producer': 21, 'ReLU': 7, 'AvgPooling2D': 1, 'Reshape': 1, 'Add': 4, 'MatMul': 1, 'Softmax': 1, 'PaddedConv2D': 7, 'Transpose': 1, 'Conv2D': 2}), defaultdict(<class 'int'>, {}))

===============\Graph manipulation
===============

Remove flatten
Fuse batchnorm
Expand metaop
Fuse to metaop

===============
New Aidge Graph
===============

  Node(name='model_batch_normalization_4_FusedBatchNormV3;model_conv2d_4_BiasAdd_ReadVariableOp_resource;model_conv2d_4_BiasAdd', optype='Producer', children: [[1]])
  Node(name='model_conv2d_8_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_8_BiasAdd_ReadVariableOp_resource1', optype='Conv2D', parents: [1, 1, 1], children: [[1]])
  Node(name='model_conv2d_5_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_5_BiasAdd_ReadVariableOp_resource1', optype='Conv2D', parents: [1, 1, 1], children: [[1]])
  Node(name='model_conv2d_5_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_5_BiasAdd_ReadVariableOp_resource', optype='Producer', children: [[1]])
  Node(name='', optype='PadConvAct', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='const_fold_opt__107', optype='Producer', children: [[1]])
  Node(name='model_activation_5_Relu;model_batch_normalization_5_FusedBatchNormV3;model_conv2d_6_BiasAdd_ReadVariableOp_resource;model_conv2d_6_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_6_Conv2D', optype='Producer', children: [[1]])
  Node(name='model_activation_Relu;model_batch_normalization_FusedBatchNormV3;model_conv2d_BiasAdd_ReadVariableOp_resource;model_conv2d_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_Conv2D', optype='Producer', children: [[1]])
  Node(name='model_activation_1_Relu;model_batch_normalization_1_FusedBatchNormV3;model_conv2d_1_BiasAdd_ReadVariableOp_resource;model_conv2d_1_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_1_Conv2D', optype='Producer', children: [[1]])
  Node(name='model_batch_normalization_2_FusedBatchNormV3;model_conv2d_2_BiasAdd_ReadVariableOp_resource;model_conv2d_2_BiasAdd', optype='Producer', children: [[1]])
  Node(name='model_batch_normalization_6_FusedBatchNormV3;model_conv2d_7_BiasAdd_ReadVariableOp_resource;model_conv2d_7_BiasAdd', optype='Producer', children: [[1]])
  Node(name='model_dense_BiasAdd_ReadVariableOp_resource', optype='Producer', children: [[1]])
  Node(name='model_activation_3_Relu;model_batch_normalization_3_FusedBatchNormV3;model_conv2d_3_BiasAdd_ReadVariableOp_resource;model_conv2d_3_BiasAdd;model_conv2d_5_Conv2D;model_conv2d_3_Conv2D', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__121', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__120', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__116', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__124', optype='Producer', children: [[1]])
  Node(name='model_conv2d_8_BiasAdd;model_conv2d_8_Conv2D;model_conv2d_8_BiasAdd_ReadVariableOp_resource', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__117', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__108', optype='Producer', children: [[1]])
  Node(name='', optype='PadConvAct', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='', optype='PadConvAct', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='const_fold_opt__113', optype='Producer', children: [[1]])
  Node(name='', optype='AddAct', parents: [1, 1], children: [[1, 1]])
  Node(name='', optype='AddAct', parents: [1, 1], children: [[1]])
  Node(name='', optype='PadConv', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='', optype='AddAct', parents: [1, 1], children: [[1, 1]])
  Node(name='', optype='PadConv', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='', optype='PadConv', parents: [1, 1, 1, 0, 0, 0], children: [[1]])
  Node(name='model_flatten_Reshape', optype='Reshape', parents: [1, 1], children: [[1]])
  Node(name='model_average_pooling2d_AvgPool', optype='AvgPooling2D', parents: [1], children: [[1]])
  Node(name='model_activation_Relu;model_batch_normalization_FusedBatchNormV3;model_conv2d_BiasAdd_ReadVariableOp_resource;model_conv2d_BiasAdd;model_conv2d_2_Conv2D;model_conv2d_Conv2D1__31', optype='Transpose', parents: [0], children: [[1]])
  Node(name='Add__29', optype='Add', parents: [1, 1], children: [[1]])
  Node(name='const_fold_opt__111', optype='Producer', children: [[1]])
  Node(name='Identity', optype='Softmax', parents: [1], children: [[]])
  Node(name='const_fold_opt__123', optype='Producer', children: [[1]])
  Node(name='const_fold_opt__125', optype='Producer', children: [[1]])
  Node(name='', optype='PadConvAct', parents: [1, 1, 1, 0, 0, 0], children: [[1, 1]])
  Node(name='model_dense_MatMul;model_dense_BiasAdd', optype='MatMul', parents: [1, 1], children: [[1]])

===============
Supported nodes 2
===============

Native operators: 39 (11 types)
- Add: 1
- AddAct: 3
- AvgPooling2D: 1
- Conv2D: 2
- MatMul: 1
- PadConv: 3
- PadConvAct: 4
- Producer: 21
- Reshape: 1
- Softmax: 1
- Transpose: 1
Generic operators: 0 (0 types)
Native types coverage: 100.0% (11/11)
Native operators coverage: 100.0% (39/39)
  (defaultdict(<class 'int'>, {'Add': 1, 'PadConv': 3, 'AddAct': 3, 'Softmax': 1, 'Conv2D': 2, 'PadConvAct': 4, 'Producer': 21, 'Transpose': 1, 'Reshape': 1, 'AvgPooling2D': 1, 'MatMul': 1}), defaultdict(<class 'int'>, {}))

===============
Supported nodes
===============

Native operators: 39 (11 types)
- Add: 1
- AddAct: 3
- AvgPooling2D: 1
- Conv2D: 2
- MatMul: 1
- PadConv: 3
- PadConvAct: 4
- Producer: 21
- Reshape: 1
- Softmax: 1
- Transpose: 1
Generic operators: 0 (0 types)
Native types coverage: 100.0% (11/11)
Native operators coverage: 100.0% (39/39)
  (defaultdict(<class 'int'>, {'PadConvAct': 4, 'Producer': 21, 'Transpose': 1, 'AddAct': 3, 'PadConv': 3, 'Reshape': 1, 'AvgPooling2D': 1, 'Add': 1, 'Softmax': 1, 'MatMul': 1, 'Conv2D': 2}), defaultdict(<class 'int'>, {}))

===============
Compile
===============

OK

===============
Create Scheduler
===============

OK

===============
Name nodes
===============

_PadConv_1_weights (Producer)
_PadConvAct_1_weights (Producer)
_PadConvAct_3_biases (Producer)
_Conv2D_0_weights (Producer)
_PadConvAct_2_biases (Producer)
_PadConvAct_2_weights (Producer)
_PadConv_1_biases (Producer)
_PadConv_2_weights (Producer)
const_fold_opt__113 (Producer)
const_fold_opt__108 (Producer)
_Conv2D_1_weights (Producer)
_PadConvAct_3_weights (Producer)
_PadConv_0_weights (Producer)
_PadConvAct_0_weights (Producer)
_PadConv_2_biases (Producer)
_Conv2D_0_biases (Producer)
_Conv2D_1_biases (Producer)
model_dense_BiasAdd_ReadVariableOp_resource (Producer)
_PadConv_0_biases (Producer)
_PadConvAct_0_biases (Producer)
_PadConvAct_1_biases (Producer)
_Transpose_0 (Transpose)
_PadConvAct_0 (PadConvAct)
_PadConvAct_1 (PadConvAct)
_PadConv_0 (PadConv)
_AddAct_0 (AddAct)
_Conv2D_0 (Conv2D)
_PadConvAct_2 (PadConvAct)
_PadConv_1 (PadConv)
_AddAct_1 (AddAct)
_Conv2D_1 (Conv2D)
_PadConvAct_3 (PadConvAct)
_PadConv_2 (PadConv)
_AddAct_2 (AddAct)
_AvgPooling2D_0 (AvgPooling2D)
_Reshape_0 (Reshape)
_MatMul_0 (MatMul)
_Add_0 (Add)
_Softmax_0 (Softmax)

===============
Set backend
===============

OK

===============
Set data format to NHWC if needed
===============

Is Layout fragil: True
⚠️ Keeping model in NCHW (layout fragile)
OK

===============
Forward dims
===============


===============
Regenerate scheduler
===============

OK

===============
Export model
===============

[[94mNOTICE[0m] - Reshape_Op: ignoring non-empty Shape attribute because input#1 takes precedence
[[33mWARNING[0m] - Reshape_Op: unable to forwardDims() because output dims are data dependent on input#1
[[94mNOTICE[0m] - Reshape_Op: ignoring non-empty Shape attribute because input#1 takes precedence
[[33mWARNING[0m] - Reshape_Op: unable to forwardDims() because output dims are data dependent on input#1
[[33mWARNING[0m] - Unable to forward dimensions (circular dependency and/or wrong dimensions and/or data
[[33mWARNING[0m]   dependent dimension?). Unable to compute output dims for nodes
[[33mWARNING[0m]   ["model_dense_MatMul;model_dense_BiasAdd (MatMul)", "Add__29 (Add)",
[[33mWARNING[0m]   "model_flatten_Reshape (Reshape)"].
[[31mERROR[0m] - Wrong input size ([1, 32, 18, 34]) for Conv operator. Expected dims are [x, 16, x, x].
[[31mERROR[0m] - Wrong input size ([1, 32, 18, 34]) for Conv operator. Expected dims are [x, 16, x, x].
[[33mWARNING[0m] - Unable to forward dimensions (circular dependency and/or wrong dimensions and/or data
[[33mWARNING[0m]   dependent dimension?). Unable to compute output dims for nodes [" (Conv2D)"].
[[33mWARNING[0m] - Unable to forward dimensions (circular dependency and/or wrong dimensions and/or data
[[33mWARNING[0m]   dependent dimension?). Unable to compute output dims for nodes [" (PadConv)"].
[[33mWARNING[0m] - Unable to forward dimensions (circular dependency and/or wrong dimensions and/or data
[[33mWARNING[0m]   dependent dimension?). Unable to compute output dims for nodes ["_PadConv_2 (PadConv)",
[[33mWARNING[0m]   "_PadConv_0 (PadConv)", "_PadConv_1 (PadConv)", "_Conv2D_1 (Conv2D)", "_Conv2D_0
[[33mWARNING[0m]   (Conv2D)", "_PadConvAct_2 (PadConvAct)", "_MatMul_0 (MatMul)", "_Add_0 (Add)",
[[33mWARNING[0m]   "_Reshape_0 (Reshape)", "_PadConvAct_1 (PadConvAct)", "_AddAct_0 (AddAct)",
[[33mWARNING[0m]   "_PadConvAct_3 (PadConvAct)"].
[[31mERROR[0m] - Assertion failed: requiredSize.type == Elts_t::Data in /opt/aidge/aidge/aidge_core/src/scheduler/SequentialScheduler.cpp:786
[[95mFATAL[0m] - Cannot generate memory with token-based producer-consumer model for node _PadConvAct_1__3
[[95mFATAL[0m]   (of type Transpose). You may need to forward dimensions in the graph first.
Traceback (most recent call last):
  File "/app/AI_Project/01_generate_cpp.py", line 169, in <module>
    aidge_core.export_utils.scheduler_export(
  File "/opt/venv/lib/python3.10/site-packages/aidge_core/export_utils/scheduler_export.py", line 79, in scheduler_export
    peak_mem, mem_info = memory_manager(scheduler, **memory_manager_args)
  File "/opt/venv/lib/python3.10/site-packages/aidge_core/mem_info.py", line 306, in generate_optimized_memory_info
    mem_manager = scheduler.generate_memory(
RuntimeError: Cannot generate memory with token-based producer-consumer model for node _PadConvAct_1__3 (of type Transpose). You may need to forward dimensions in the graph first.
Error: export_model/data directory does not exist.

error.log

sed: can't read export_model/Makefile: No such file or directory
sed: can't read export_model/Makefile: No such file or directory
sed: can't read export_model/Makefile: No such file or directory
sed: can't read export_model/Makefile: No such file or directory
make: AI_Build AI_Deploy AI_Manager AI_Project AI_Support ConvNet.onnx MLP_MNIST.onnx MobileNet-v2.onnx README.md Resnet.onnx __pycache__ config.json docker examples exit_functions.sh export_model mnist.onnx mnist_test_input_type2.bin model_1D_classifier.onnx model_type2.onnx print_raw_output.py type1_test.sh type2_test.sh type3_test.sh No targets specified and no makefile found. Stop.

Report Details

report.json

null

214335 - rpi5 aidge 214337 - rpi5 aidge