Data-Centric and Heterogeneity-Adaptive Sequence Parallelism for Efficient LLM Training

Add code
Dec 02, 2024

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: