Skip to content

[Feat] remove flatten in atom sglang mla like atom vllm mla #16

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

Open
ZLkanyo009 wants to merge 1 commit into
base: plugin_for_sgl_oot
Choose a base branch
from
Open

[Feat] remove flatten in atom sglang mla like atom vllm mla #16

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
41 changes: 39 additions & 2 deletions atom/plugin/sglang/attention_backend/sgl_attention_mla.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -236,6 +236,43 @@ def mla_absorbed_bmm(
return torch.bmm(inp.transpose(0, 1), weight).transpose(0, 1)


def mla_v_up_proj(
attn: DeepseekV2MLAAttention,
inp: torch.Tensor,
weight: torch.Tensor,
weight_scale: Optional[torch.Tensor],
weight_scale_k: Optional[torch.Tensor],
out_dim: int,
) -> torch.Tensor:
"""Project MLA decode output to a flat o_proj input."""
if _is_hip and (
(_use_aiter_gfx95 and weight.dtype == torch.float8_e4m3fn)
or (get_is_capture_mode() and weight.dtype == torch.float8_e4m3fnuz)
):
x = inp.transpose(0, 1)
out = torch.empty(
(inp.shape[0], attn.num_local_heads * out_dim),
device=inp.device,
dtype=torch.bfloat16,
)
out_3d = out.view(inp.shape[0], attn.num_local_heads, out_dim)
batched_gemm_a8w8_a_per_token_group_prequant_w_per_batched_tensor_quant(
X=x,
WQ=weight.transpose(-1, -2),
w_scale=weight_scale,
group_size=128,
YQ=out_3d,
transpose_bm=True,
transpose_bm_in=False,
dtype=torch.bfloat16,
)
return out

return mla_absorbed_bmm(
attn, inp, weight, weight_scale, weight_scale_k, out_dim
).flatten(1, 2)


# Forward: prepare → core
def forward_sgl_prepare(
attn: DeepseekV2MLAAttention,
Expand Down Expand Up @@ -420,9 +457,9 @@ def forward_sgl_core(
attn_output = attn_output.view(-1, attn.num_local_heads, attn.kv_lora_rank)

# up-proj by w_vc
attn_bmm_output = mla_absorbed_bmm(
attn_bmm_output = mla_v_up_proj(
attn, attn_output, attn.w_vc, attn.w_scale, attn.w_scale_v, attn.v_head_dim
).flatten(1, 2)
)

return attn.o_proj(attn_bmm_output)

Expand Down