Added flash attention for speedup
04608fc
Sualeh Qureshi
commited on
Commit History
Added time logs of per batch and tokens per second for testing speedups performance
34ef2b4
Sualeh Qureshi
commited on
Started working on Optimization / speed ups. in this commit implemented scaled weight initialization
d830e2d
Sualeh Qureshi
commited on
Commited the training code and model file
c175ce3
Sualeh Qureshi
commited on