Skip to content

[FIX_FOR_VLLM_CUSTOM=b06813e87207e15b133e903d641e03f237d85b17] Fix gdn_linear_attn import path after upstream mamba refactor #1496

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
adobrzyn merged 1 commit into from
May 27, 2026
Merged

[FIX_FOR_VLLM_CUSTOM=b06813e87207e15b133e903d641e03f237d85b17] Fix gdn_linear_attn import path after upstream mamba refactor #1496

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
12 changes: 6 additions & 6 deletions vllm_gaudi/models/qwen3_5.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
import torch
from vllm.model_executor.layers.mamba.gdn_linear_attn import GatedDeltaNetAttention
from vllm.model_executor.layers.mamba.gdn.qwen_gdn_linear_attn import QwenGatedDeltaNetAttention
from vllm.forward_context import get_forward_context

from vllm_gaudi.ops.causal_conv1d_pytorch import (
Expand All @@ -26,7 +26,7 @@ def _save_ssm_state(core_attn_out, final_state, ssm_state, state_indices):
return core_attn_out


class HPUGatedDeltaNetAttention(GatedDeltaNetAttention):
class HPUGatedDeltaNetAttention(QwenGatedDeltaNetAttention):

def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
Expand Down Expand Up @@ -295,10 +295,10 @@ def forward(

# Replace the class in the upstream modules so that both Qwen3-Next and
# Qwen3.5 model definitions instantiate HPUGatedDeltaNetAttention.
import vllm.model_executor.layers.mamba.gdn_linear_attn as _gdn_module # noqa: E402
import vllm.model_executor.layers.mamba.gdn.qwen_gdn_linear_attn as _gdn_module # noqa: E402
import vllm.model_executor.models.qwen3_next as _qwen3_next_module # noqa: E402
import vllm.model_executor.models.qwen3_5 as _qwen3_5_module # noqa: E402

_gdn_module.GatedDeltaNetAttention = HPUGatedDeltaNetAttention
_qwen3_next_module.GatedDeltaNetAttention = HPUGatedDeltaNetAttention
_qwen3_5_module.GatedDeltaNetAttention = HPUGatedDeltaNetAttention
_gdn_module.QwenGatedDeltaNetAttention = HPUGatedDeltaNetAttention
_qwen3_next_module.QwenGatedDeltaNetAttention = HPUGatedDeltaNetAttention
_qwen3_5_module.QwenGatedDeltaNetAttention = HPUGatedDeltaNetAttention
Loading