search

microsoft / MMdnn

MMdnn is a set of tools to help users inter-operate among different deep learning frameworks. E.g. model conversion and visualization. Convert models between Caffe, Keras, MXNet, Tensorflow, CNTK, PyTorch Onnx and CoreML.
MIT License
5.78k stars 968 forks source link

no npy weights file keras to pytorch #903

Closed dasturge closed 3 years ago

dasturge commented 3 years ago

Platform (like ubuntu 16.04/win10): ubuntu 20.04 Python version: 3.6 Source framework with version (like Tensorflow 1.4.1 with GPU): keras 2.2.4, tensorflow-gpu 1.14 Destination framework with version (like CNTK 2.3 with GPU): pytorch 1.4.0 Pre-trained model path (webpath or webdisk path): model is a transfer learning resnet50 Running scripts: mmconvert -sf keras -df pytorch -in custom_resnet50.json -iw custom-resnet50-weights.h5 -om weights.pt

so the command outputs a file weights.py and weights.pt, and says something about a "b707ee71b00a45dd8d4c2fd686a7c643.npy" weights file being saved, but the weights file is not actually saved in the working directory and yet is necessary to actually load the pytorch model which is generated.

cookieli commented 3 years ago

Thank you for your issues. I'm sorry for repling too late. I think if mmdnn reports no errors otherwise. Then weights file should be generated correctly. Maybe you should provide more details.