(1)create lenet modle: create_model.py and create_seumodel.py. create_seumodel.py is intended for testing the radiation-hardened capability of a modified training method or altered network structure.However, it is currently not yet operational.
(2) fault injection : fault_injection.py seu.py
(3) Top-level-testing : seutest.py seutest_noseu.py seutest2.py seutest3.py. seutest_noseu also cannot be used, for same reason as create_seumodel.
(4) Data processing : dealdata.py extem_p.py get_model_extrem_num.py weights_distribution.py. In fact, much of our data processing is done directly in the top-level testing. Here is some code for obtaining the probabilities mentioned in the paper.
(5) Results and others : The remaining
minghan1/Letnet-seu
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|