Skip to content

[shardformer] fix linear 1d row and support uneven splits for fused qkv linear #6084

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

Merged
ver217 merged 5 commits into from
Oct 10, 2024
Merged

[shardformer] fix linear 1d row and support uneven splits for fused qkv linear #6084

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
d46c8b2
[tp] hotfix linear row
ver217 Oct 9, 2024
e78689f
[tp] support uneven split for fused linear
ver217 Oct 9, 2024
71f6c14
[tp] support sp for fused linear
ver217 Oct 9, 2024
b6e5f9d
[tp] fix gpt2 mlp policy
ver217 Oct 9, 2024
1d72639
[tp] fix gather fused and add fused linear row
ver217 Oct 10, 2024
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
4 changes: 3 additions & 1 deletion colossalai/inference/modeling/policy/nopadding_baichuan.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,9 @@ def module_policy(self):
target_module=NopadBaichuanMLP,
),
SubModuleReplacementDescription(
suffix="self_attn.W_pack", target_module=FusedLinear1D_Col, kwargs={"n_fused": 3}
suffix="self_attn.W_pack",
target_module=FusedLinear1D_Col,
kwargs={"split_sizes": [self.model.config.hidden_size] * 3},
),
SubModuleReplacementDescription(
suffix="self_attn.o_proj",
Expand Down
3 changes: 2 additions & 1 deletion colossalai/shardformer/layer/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
from .loss import cross_entropy_1d, dist_cross_entropy
from .normalization import FusedLayerNorm, FusedRMSNorm, LayerNorm, RMSNorm
from .parallel_module import ParallelModule
from .qkv_fused_linear import FusedLinear1D_Col, GPT2FusedLinearConv1D_Col, GPT2FusedLinearConv1D_Row
from .qkv_fused_linear import FusedLinear1D_Col, FusedLinear1D_Row, GPT2FusedLinearConv1D_Col, GPT2FusedLinearConv1D_Row

__all__ = [
"Embedding1D",
Expand Down Expand Up @@ -34,4 +34,5 @@
"RingAttention",
"get_pad_info",
"all_to_all_comm",
"FusedLinear1D_Row",
]
4 changes: 2 additions & 2 deletions colossalai/shardformer/layer/_operation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -840,7 +840,7 @@ def forward(ctx, input_, process_group, scatter_dim, gather_dim, fp8_communicati
ctx.gather_dim = gather_dim
ctx.fp8_communication = fp8_communication
world_size = dist.get_world_size(process_group)
bsz, _, _ = input_.shape
bsz = input_.shape[0]

# using all_to_all_single when batch size is 1
if bsz == 1:
Expand Down Expand Up @@ -871,7 +871,7 @@ def backward(ctx, grad_output):
gather_dim = ctx.scatter_dim
fp8_communication = ctx.fp8_communication
world_size = dist.get_world_size(process_group)
bsz, _, _ = grad_output.shape
bsz = grad_output.shape[0]

if bsz == 1:
return_grad = _all_to_all_single(
Expand Down
11 changes: 4 additions & 7 deletions colossalai/shardformer/layer/linear.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -428,11 +428,8 @@ def forward(self, input_: Tensor) -> Tensor:
handle.wait()
output = torch.cat(output_parallel_list, dim=-1)
else:
if self.seq_parallel_mode is None:
output_parallel = linear_with_async_comm(input_, self.weight, None, self.process_group, False)
output = reduce_forward(output_parallel, self.process_group, fp8_communication=self.fp8_communication)
elif self.seq_parallel_mode == "split_gather":
output_parallel = linear_with_async_comm(input_, self.weight, None, self.process_group, False)
if self.seq_parallel_mode == "split_gather":
output_parallel = F.linear(input_, self.weight)
output = reducescatter_forward_gather_backward(
output_parallel, self.process_group, self.seq_parallel_dim, fp8_communication=self.fp8_communication
)
Expand All @@ -445,8 +442,8 @@ def forward(self, input_: Tensor) -> Tensor:
ring=True,
)
else:
output_parallel = linear_with_async_comm(input_, self.weight, None, self.process_group, False)
output = reduce_forward(output_parallel, self.process_group)
output_parallel = F.linear(input_, self.weight)
output = reduce_forward(output_parallel, self.process_group, fp8_communication=self.fp8_communication)

if not self.skip_bias_add:
if self.bias is not None:
Expand Down
Loading