carter54

Results 17 comments of carter54

[Numpy Refactor] [Model Deployment] Use TVM to accelerate model inference + deployment

any tutorial to deploy gluon gpt-2 model with TVM?

[Numpy Refactor] [Model Deployment] Use TVM to accelerate model inference + deployment

Thanks @sxjscience. Looking forward to try it~

[Numpy Refactor] [Model Deployment] Use TVM to accelerate model inference + deployment

@sxjscience Thanks a lot! I will have a try

Difference between embeddings Gluon and Huggingface

@devsentient @evah88 I'm also trying to transfer model from huggingface to mxnet by matching parameter names, but I am transfering GPT 2. Initially I also got different results from hugginface...

bolt编译后能在visual studio导入使用吗？

![image](https://user-images.githubusercontent.com/26741594/136488314-fbb378d9-770f-4820-b1ac-0d4d3b58ad0b.png) 导入的时候报了一些错。。

bolt编译后能在visual studio导入使用吗？

@yuxianzhi 多谢回复~ 如果用bolt.h，是指代码里的c++的API都要换成C的API吗？ 比如creatPipeline要换成CreateModel ？

bolt编译后能在visual studio导入使用吗？

@yuxianzhi 好的 谢谢~

mxnet.base.MXNetError: Failed loading Op yolo_output of type _contrib_Yolo2Output: [13:11:48] src/core/op.cc:55: Check failed: op != nullptr Operator _contrib_Yolo2Output is not registered

@jacky4323 I'm facing the same issue... Check failed: op != nullptr Operator stack_neighbor is not registered

mxnet.base.MXNetError: Failed loading Op yolo_output of type _contrib_Yolo2Output: [13:11:48] src/core/op.cc:55: Check failed: op != nullptr Operator _contrib_Yolo2Output is not registered

It seems only the yolo branch mxnet has stack_neighbor in src/operator/tensor/matrix_op.xx , while the master branch only has stack/stack_op

How to freeze checkpoint graph to .pb format?

output.name gives you a tensor name (' sample_sequence/while/Exit_3:0') not a node name. I guess you should put [ 'sample_sequence/while/Exit_3'] as argument 3 in tf.graph_util.convert_variables_to_constants