rtdetr进行自动压缩过程中报错如下

[desinerPol]

运行指令: python run.py --config_path=./configs/rtdetr_hgnetv2_x_qat_dis.yaml --save_dir='./output/' --devices='cpu'

配置文件：

Global: reader_config: configs/rtdetr_reader.yml include_nms: False Evaluation: True model_dir: ./models/rtdetr_hgnetv2_6x_coco/ model_filename: model.pdmodel params_filename: model.pdiparams

Distillation: alpha: 1.0 loss: soft_label

TrainConfig: train_iter: 500 eval_iter: 100 learning_rate: type: CosineAnnealingDecay learning_rate: 0.00003 T_max: 10000 optimizer_builder: optimizer: type: SGD weight_decay: 4.0e-05

报错信息： C++ Traceback (most recent call last):

0 paddle::framework::ThreadPoolTemplpaddle::framework::StlThreadEnvironment::WorkerLoop(int) 1 paddle::framework::ProgramInterpreter::RunInstructionAsync(unsigned long) 2 paddle::framework::ProgramInterpreter::RunInstruction(paddle::framework::Instruction const&) 3 paddle::framework::ProgramInterpreter::RunOperator(paddle::framework::Instruction const&) 4 phi::KernelImpl<void ()(phi::CPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, std::string const&, std::string const&, bool, phi::DenseTensor), &(void phi::GridSampleKernel<float, phi::CPUContext>(phi::CPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, std::string const&, std::string const&, bool, phi::DenseTensor*))>::Compute(phi::KernelContext*) 5 void phi::GridSampleKernel<float, phi::CPUContext>(phi::CPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, std::string const&, std::string const&, bool, phi::DenseTensor*)

[wanghaoshuang]

请问您使用的 PaddlePaddle 和 PaddleSlim 版本格式多少？

[desinerPol]

请问您使用的 PaddlePaddle 和 PaddleSlim 版本格式多少？

paddlepaddle-gpu 2.6.1 paddleSlim git develop