Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[format] applied code formatting on changed files in PR 5510 #5517

Merged
merged 1 commit into from
Mar 27, 2024
Merged

[format] applied code formatting on changed files in PR 5510 #5517

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion colossalai/shardformer/modeling/gpt2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1302,7 +1302,6 @@ def forward(
shift_logits, shift_labels, process_group=shard_config.tensor_parallel_process_group
)


if not shard_config.parallel_output:
lm_logits = gather_forward_split_backward(lm_logits, -1, shard_config.tensor_parallel_process_group)

Expand Down
2 changes: 0 additions & 2 deletions colossalai/shardformer/modeling/llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,10 +15,8 @@
from colossalai.pipeline.stage_manager import PipelineStageManager
from colossalai.shardformer.shard import ShardConfig


from ..layer import ColoAttention, cross_entropy_1d


try:
from transformers.models.llama.modeling_llama import _prepare_4d_causal_attention_mask

Expand Down
8 changes: 6 additions & 2 deletions colossalai/shardformer/policies/gpt2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -291,13 +291,17 @@ def module_policy(self):
GPT2LMHeadModel: ModulePolicyDescription(
sub_module_replacement=[
SubModuleReplacementDescription(
suffix="lm_head", target_module=col_nn.Linear1D_Col, kwargs={"gather_output": not self.shard_config.parallel_output}
suffix="lm_head",
target_module=col_nn.Linear1D_Col,
kwargs={"gather_output": not self.shard_config.parallel_output},
)
],
)
}
if self.shard_config.parallel_output:
addon_module[GPT2LMHeadModel].method_replacement={"forward": get_lm_forward_with_dist_cross_entropy(self.shard_config)}
addon_module[GPT2LMHeadModel].method_replacement = {
"forward": get_lm_forward_with_dist_cross_entropy(self.shard_config)
}
module_policy.update(addon_module)

if self.pipeline_stage_manager is not None:
Expand Down
10 changes: 8 additions & 2 deletions colossalai/shardformer/policies/llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -265,12 +265,18 @@ def module_policy(self):
new_item = {
LlamaForCausalLM: ModulePolicyDescription(
sub_module_replacement=[
SubModuleReplacementDescription(suffix="lm_head", target_module=Linear1D_Col, kwargs={"gather_output": not self.shard_config.parallel_output})
SubModuleReplacementDescription(
suffix="lm_head",
target_module=Linear1D_Col,
kwargs={"gather_output": not self.shard_config.parallel_output},
)
],
)
}
if self.shard_config.parallel_output:
new_item[LlamaForCausalLM].method_replacement={"forward": get_lm_forward_with_dist_cross_entropy(self.shard_config)}
new_item[LlamaForCausalLM].method_replacement = {
"forward": get_lm_forward_with_dist_cross_entropy(self.shard_config)
}
policy.update(new_item)

if self.pipeline_stage_manager:
Expand Down