PlantCPU.ipynb
CUDA was throwing error so i started working with cpu and it takes very long time to execute
I knew that cpu is not gonna complete the task so i have to use GPU
Plant.ipynb
Then i installed CUDA but still got the error that torch is not compiled with CUDA
then i thought of using the CUDA and CUdnn for the pytorch installed and completed installation process Finally it was working
the first step was executed with so less time compared to cpu process i got hopes
and then this happened
i was getting same error after running multiple time
then i started looking how to tackle with this problem
i came across the idea to reduce the batch size
i reduced the size to 28 from 32 then 24, 20, 16 and finally 8
finally it started running even though it used almost 3.9gb out of 4gb still it executed after long time
after waiting for some more it finally completed
finally it was executed with accuracy of 97.52
but we need accuracy of more than 99.2