EthanYe
EthanYe
### PR types Bug fixes ### PR changes OPs ### Describe Fix reorder bug in conv MKLDNN
### PR types Bug fixes ### PR changes Others ### Describe When the interpolate OP only specifies the output dim but does not specify the scale, the value of the...
### PR types Others ### PR changes Others ### Description Use paddle2onnx.legacy for lower version of ONNXRuntime-gpu
1. Set input spec to dynamic by default 2. If you don't need to set dynamic shape, set input_spec_shape to None 3. fix lookup_table_v2 and gaussian_random dynamic shape bugs
支持Paddle量化模型导出 1. 如果你导出的ONNX模型用TensorRT进行部署,TensorRT可直接加载该模型做量化推理 ``` paddle2onnx --model_dir quant_inference_model/ --model_filename model.pdmodel --params_filename model.pdiparams --save_file model.onnx --opset_version 13 --enable_onnx_checker True --deploy_backend TensorRT ``` 2. 如果你导出的ONNX模型用ONNXRuntime进行部署,ONNXRuntime可直接加载该模型做量化推理 ``` paddle2onnx --model_dir quant_inference_model/ --model_filename model.pdmodel --params_filename model.pdiparams...
Reopen quantize related op test
1. Add cpp_demo for linux
1. add box_clip test and fix one bug 2. change custom domain from custom to baidu
1. 增加Selected_row数据输入格式的支持。 2. 增加sequence_pool op转换支持及其单测,但是LD Tensor的Lod level为1且所有lod的只有一个数字,为batch维度,或者batch=1
### PR types Bug fixes ### PR changes Others ### Describe Remove calibration file path when deploy quantize model