Demystifying Workload Imbalances in Large Transformer Model Training over Variable-length Sequences

Add code
Dec 10, 2024

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: