DeepSpeed
add lm_head and embed_out tensor parallel
#3962
Merged

add lm_head and embed_out tensor parallel #3962

Yejing-Lai
Yejing-Lai Yejing-Lai requested a review from RezaYazdaniAminabadi RezaYazdaniAminabadi 2 years ago
Yejing-Lai Yejing-Lai requested a review from jeffra jeffra 2 years ago
Yejing-Lai Yejing-Lai requested a review from mrwyattii mrwyattii 2 years ago
Yejing-Lai Yejing-Lai requested a review from awan-10 awan-10 2 years ago
Yejing-Lai Yejing-Lai requested a review from cmikeh2 cmikeh2 2 years ago
Yejing-Lai Yejing-Lai requested a review from arashb arashb 2 years ago
RezaYazdaniAminabadi
RezaYazdaniAminabadi RezaYazdaniAminabadi closed this 2 years ago
RezaYazdaniAminabadi RezaYazdaniAminabadi reopened this 2 years ago
Yejing-Lai
delock
delock commented on 2023-07-24
delock
delock commented on 2023-07-24
delock
delock
dc3671 dc3671 force pushed from 3036650d to 92303cb2 2 years ago
dc3671
dc3671 commented on 2023-08-16
dc3671
dc3671
RezaYazdaniAminabadi
RezaYazdaniAminabadi approved these changes on 2023-09-29
RezaYazdaniAminabadi RezaYazdaniAminabadi enabled auto-merge 2 years ago
disabled auto-merge 2 years ago
Manually disabled by user
dc3671 dc3671 force pushed from ea7fb280 to 10fbdbc3 2 years ago
dc3671
Yejing-Lai add lm_head and embed_out tensor parallel
fecbfdd8
Yejing-Lai fix load lm_head.weight name issue
3ef32311
Yejing-Lai replace all_reduce with inference_all_reduce
e3e60e5b
refactor lm_head tensor parallel
fc64ef5b
dc3671 dc3671 force pushed from ecc676ad to fc64ef5b 2 years ago
dc3671
tjruwase tjruwase merged 6763e2de into master 2 years ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone