Skip to content

fix: fix accuracy bug in flashinfer and fa3 kernel. #995

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
hiworldwzj merged 3 commits into from
Aug 4, 2025
Merged

fix: fix accuracy bug in flashinfer and fa3 kernel. #995

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
087ca04
fix: fix a bug in flashinfer
Aug 1, 2025
fbf3459
fix: fix a bug in fa3 in deepseek2
Aug 4, 2025
1348186
fix
hiworldwzj Aug 4, 2025
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
7 changes: 4 additions & 3 deletions lightllm/models/deepseek2/layer_infer/transformer_layer_infer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@
from lightllm.models.deepseek2.triton_kernel.rotary_emb import rotary_emb_fwd
from lightllm.models.deepseek2.infer_struct import Deepseek2InferStateInfo
from lightllm.models.deepseek2.flashinfer_struct import Deepseek2FlashInferStateInfo
from lightllm.models.deepseek2.flashattention_infer_struct import Deepseek2FlashAttentionStateInfo
from functools import partial
from lightllm.models.llama.yarn_rotary_utils import get_deepseek_mscale
from lightllm.distributed.communication_op import all_gather, all_gather_into_tensor, all_reduce, reduce_scatter_tensor
Expand Down Expand Up @@ -302,7 +303,7 @@ def _context_attention_flashattention_kernel_with_CC(
self,
q: torch.Tensor,
kv,
infer_state: Deepseek2FlashInferStateInfo,
infer_state: Deepseek2FlashAttentionStateInfo,
layer_weight: Deepseek2TransformerLayerWeight,
out=None,
) -> torch.Tensor:
Expand All @@ -323,7 +324,7 @@ def _context_attention_flashattention_kernel_with_CC(
k=k.view(-1, self.tp_k_head_num_, self.qk_nope_head_dim + self.qk_rope_head_dim),
v=v.view(-1, self.tp_v_head_num_, self.v_head_dim),
cu_seqlens_q=infer_state.cu_seqlens_q,
cu_seqlens_k=infer_state.cu_seqlens_k,
cu_seqlens_k=infer_state.cu_seqlens_q,
max_seqlen_q=infer_state.q_max_seq_len,
max_seqlen_k=infer_state.max_seq_len,
softmax_scale=self.softmax_scale,
Expand Down Expand Up @@ -547,7 +548,7 @@ def _context_attention_kernel_origin_fp8(
return o_tensor

def _token_gqa_decode_attention_flashattention(
self, q, infer_state: Deepseek2FlashInferStateInfo, layer_weight: Deepseek2TransformerLayerWeight, out=None
self, q, infer_state: Deepseek2FlashAttentionStateInfo, layer_weight: Deepseek2TransformerLayerWeight, out=None
):
q_nope, q_rope = q[:, :, : -self.qk_rope_head_dim], q[:, :, -self.qk_rope_head_dim :]
q_nope = layer_weight.k_b_proj_.bmm(q_nope.transpose(0, 1)).transpose(0, 1)
Expand Down
4 changes: 2 additions & 2 deletions lightllm/models/llama/flashinfer_struct.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,8 +81,8 @@ def init_some_extra_state(self, model, input_ids: torch.Tensor):
self.req_manager.req_to_token_indexs,
self.b_req_idx,
self.b_seq_len,
kv_starts,
self.max_len_in_batch,
kv_starts[:-1],
self.max_kv_seq_len,
kv_indices,
)
self.prefill_wrapper = flashinfer.prefill.BatchPrefillWithPagedKVCacheWrapper(
Expand Down