[Refactor] Move fla to third party (#48500)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
@@ -274,7 +274,7 @@ def _warm_causal_conv1d_fwd_kernel(
|
||||
def _warm_fused_post_conv_kernel(
|
||||
device: torch.device, config: _QwenGDNWarmupConfig
|
||||
) -> None:
|
||||
from vllm.model_executor.layers.fla.ops.fused_gdn_prefill_post_conv import (
|
||||
from vllm.third_party.flash_linear_attention.ops.fused_gdn_prefill_post_conv import ( # noqa: E501
|
||||
fused_post_conv_prep,
|
||||
)
|
||||
|
||||
@@ -304,7 +304,7 @@ def _warm_fused_sigmoid_gating_delta_rule_update_kernel(
|
||||
device: torch.device,
|
||||
config: _QwenGDNWarmupConfig,
|
||||
) -> None:
|
||||
from vllm.model_executor.layers.fla.ops.fused_sigmoid_gating import (
|
||||
from vllm.third_party.flash_linear_attention.ops.fused_sigmoid_gating import (
|
||||
fused_sigmoid_gating_delta_rule_update,
|
||||
)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user