Update Attention op to support attention mask for GPT-2 #4330
Fix mask in EmbedLayerNormalization
d41754fe
Add MaskIndex cuda op for BERT optimization
8864979a
Update dynamic axes of gpt2 with past state
51d68cf3
tianleiwu
marked this pull request as draft 5 years ago
Add attention mask for GPT-2 in cuda & cpu Attention operator
bb8e3958
Merge master
0b91f1ab
tianleiwu
marked this pull request as ready for review 5 years ago
tianleiwu
changed the title WIP: Support Attention Mask for GPT-2 with past state Update Attention op to support attention mask for GPT-2 with left side padding 5 years ago
Add back unit tests that removed in merge.
809249d9
format
eb3ed536
Remove printf
1fbf5002
Support 2D attention mask
91bdc45f
fix build warning
fdefb43a
Update script to fuse model with attention mask
0c97b26c
tianleiwu
changed the title Update Attention op to support attention mask for GPT-2 with left side padding Update Attention op to support attention mask for GPT-2 5 years ago
liuziyue
approved these changes
on 2020-06-30
tianleiwu
merged
55f25a4b
into master 5 years ago
tianleiwu
deleted the tlwu/gpt2_attention_mask branch 5 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub