make both TP overlap splitting by batch and seq work together#630

Merged

xuechendi merged 1 commit intoHabanaAI:mlperf_featuresfrom xuechendi:tp_parallelism_2_rebased

Dec 16, 2024

+180-68

Commits on Dec 16, 2024

Support tensor split in llama decoderlayer

jikunshang
authored and
xuechendi
committed