Junshen Xu

icon indicating a website link https://www.mit.edu/~junshen/ icon indicating an email [email protected]

icon company MIT icon location Cambridge icon location PhD student @ MIT EECS

Results 1 comments of Junshen Xu

Efficiency with deepspeed

Thanks for the suggestion. Actually we have already used some of those techniques, e.g., mixed precision training, to improve the usage of GPU memory and efficiency. Other techniques, such as...