/LSTMVariants.jl

MIT Numerical Computing - LSTM/GRU Variants Julia

Primary LanguageJupyter Notebook

LSTM from scratch

Task_1_6_LSTM_LAB - Notebook has the first part of the lab

Task7_LSTM_With_GPU.jl is the GPU enabled julia file you can run on a GPU. The only difference is this notebook has the global parameter that you can change to adjust array type

Task_8_analyse performance has some discussion about profiling, memory, type instabilities and some epoch results

Task_9_GRU_Variant - Has the GRU variant of LSTM.

Task_9_GRU_Variant_GPU is the corresponding GPU file

.bin files in the repo are the saved versions of the profiling of the training function .

.mem files are the memory allocation tracker files