assignment 3 completed

극강의 난이도..

극강의 분량..

 

진짜 진짜 엄청 힘들었다..

 

마지막 Batch Norm 에러 난 거 있는데.. 도저히 수정을 해도해도 debugging이 안되서.. 포기.

사실 힘이 다 빠졌다.

 

이렇게 힘든 코딩은 난생 처음이네..

 

한창 코테, PS 취미삼아 풀 때보다 더더더 너무 힘들었다 엉엉

 

나 칭찬해주세요.. ㅜㅜ 

 

It took approximately 15~16 hrs.. T.T

 

If you happen to come across my code, please just reference it only. This is my own solution, so it has not been "verified" (Actually the performance is not all good). 

Also, while it passed all unit tests, there was one test that I couldn't pass. T.T

I really tried, but I couldn't figure out how to solve it.. If you have any comment about my code, please leave a comment~!

Any comments are welcome!

 

* Implementations

- Linear layer

- Conv layer

- Max pooling

- ReLU activation

- Batch Norm

- Dropout

- Weight initialization - Kaiming (Xavier)

- Optimizers - sgd, sgd+momentum, rmsprop, adam

- FullyConnected Network

- DeepConv Network

 

https://github.com/letter-night/STA9131