search

openvinotoolkit / open_model_zoo

Pre-trained Deep Learning models and demos (high quality and extremely fast)
https://docs.openvino.ai/latest/model_zoo.html
Apache License 2.0
4.1k stars 1.38k forks source link

Annotations for bert model cannot be loaded by quantization_model_evaluator #539

Closed dliang0406 closed 5 years ago

dliang0406 commented 5 years ago

File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/evaluators/quantization_model_evaluator.py", line 99, in process_dataset_async self.select_dataset(dataset_tag) File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/evaluators/quantization_model_evaluator.py", line 167, in select_dataset dataset_attributes = create_dataset_attributes(self.dataset_config, dataset_tag) File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/evaluators/quantization_model_evaluator.py", line 339, in create_dataset_attributes data_reader = BaseReader.provide(data_reader_config['type'], data_source, data_reader_config) File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/dependency.py", line 67, in provide return root_provider(*args, **kwargs) File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/data_readers/data_reader.py", line 320, in init super().init(annotations, config) File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/data_readers/data_reader.py", line 90, in init self.configure() File "/home/dan/openvino/post-training-compression-tool/libs/open_model_zoo/tools/accuracy_checker/accuracy_checker/data_readers/data_reader.py", line 326, in configure if not contains_all(self.data_source[0].dict, self.feature_list): TypeError: 'NoneType' object is not subscriptable

eaidova commented 5 years ago

@dliang0406, I suppose we can close this issue, right?

dliang0406 commented 5 years ago

Yes, I'll close it, thank you.