Add mark step and inplace residual add in llama model code to reduce memory consumption#65

7 commits merged into

HabanaAI:habana-mainHabanaAI/optimum-habana-fork:habana-mainfrom

puneeshkhanna:llama_prefill_memoryfixespuneeshkhanna/optimum-habana-fork:llama_prefill_memoryfixesCopy head branch name to clipboard