DeepSpeed inference config. (#2459) #2472
Prototype DS inference config. Tested with gpt2/bert. (#2459)
1b0c2fb5
Merge branch 'master' into staging-inf-config-v1
0cc4fb7b
add the missing max tokens fix.
7f192540
awan-10
marked this pull request as ready for review 3 years ago
fix format.
69db92d8
Merge branch 'master' into staging-inf-config-v1
8f396386
moe fixes.
72c0b99a
Merge branch 'master' into staging-inf-config-v1
8f758b6b
Merge branch 'master' into staging-inf-config-v1
bc3af199
fix mp_size validation not properly setting tp_size
61380387
fix broken custom injection policy
36a53c9c
Merge branch 'master' into staging-inf-config-v1
c7dbb33f
Merge branch 'master' into staging-inf-config-v1
399f45b5
first round of config to be passed to apply_injection_policy. manual
ac95ab7f
fix several things.
38bb2088
format fix
b3b451c2
Merge branch 'master' into staging-inf-config-v1
cc7ccb5f
awan-10
changed the title Prototype DS inference config. Tested with gpt2/bert. (#2459) DeepSpeed inference config. (#2459) 3 years ago
Merge branch 'master' into staging-inf-config-v1
bbe351f5
cmikeh2
approved these changes
on 2022-11-14
moved docs to appropriate place in config model
266e60bd
added bf16 support for dtype
a3d70c22
mrwyattii
approved these changes
on 2022-11-14
mrwyattii
enabled auto-merge (squash) 3 years ago
mrwyattii
merged
b5d18a6a
into master 3 years ago
mrwyattii
deleted the staging-inf-config-v1 branch 2 years ago
Login to write a write a comment.
Login via GitHub