[AutoParallel] Add sequence parallel for llama #59822
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
… fix_backward_reshard
… fix_backward_reshard
… fix_backward_reshard
… fix_backward_reshard
|
你的PR提交成功，感谢你对开源项目的贡献! |
GhostScreaming
changed the title
GhostScreaming
force-pushed
the
test_sp
branch
from
5972d6f
to
d6c38d9
Compare
JZ-LIANG
reviewed
Dec 8, 2023
| @@ -192,6 +192,26 @@ def forward( | |||
| shape=target_key_value_shape | |||
| ) | |||
|
|
|||
| if self.config.sequence_parallel: | |||
| query_states = dist.reshard( | |||
There was a problem hiding this comment.
reshard before projection while tranpose after projection
JZ-LIANG
reviewed
Dec 8, 2023
| @@ -238,6 +250,12 @@ def forward( | |||
| else: | |||
| attn_output = outputs | |||
|
|
|||
| if self.config.sequence_parallel: | |||
| attn_output = paddle.transpose(attn_output, [1, 0, 2]) | |||
| attn_output = dist.reshard( | |||
There was a problem hiding this comment.
reshard should after out_projection
| @@ -731,6 +772,13 @@ def forward( | |||
|
|
|||
| # if labels is None，means we need full output, instead of tensor_parallel_output | |||
| logits = self.lm_head(hidden_states) | |||
| if self.config.sequence_parallel: | |||
There was a problem hiding this comment.
# reshard should before lm_head
if self.config.sequence_parallel:
hidden_states = dist.reshard(
hidden_states, get_mesh(-1), [dist.Shard(1), dist.Replicate()]
)
# [S, B, H] -> [B, S, H]
hidden_states = paddle.transpose(hidden_states, [1, 0, 2])
logits = self.lm_head(hidden_states)
XieYunshen
approved these changes
Dec 9, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR types
Others
PR changes
Others
Description
Pcard-73145
Add sequence parallel for llama.