[Model] Remove unused DeepSeek V4 top-k buffer helper (#53697)
This commit is contained in:
@@ -685,13 +685,6 @@ class DeepseekV4Attention(nn.Module, AttentionLayerBase, ABC):
|
||||
)
|
||||
return q_fp8
|
||||
|
||||
def _global_topk_output_buffers(
|
||||
self, topk_indices: torch.Tensor
|
||||
) -> tuple[torch.Tensor, torch.Tensor] | None:
|
||||
if self.compress_ratio != 4 or self.eager_scratch_pool is None:
|
||||
return None
|
||||
return self.eager_scratch_pool.global_topk_outputs(topk_indices)
|
||||
|
||||
def bind_kv_cache(self, kv_cache: torch.Tensor) -> None:
|
||||
# [B, H=1, N, C] -> [B, N, C]
|
||||
self.kv_cache = kv_cache.squeeze(1)
|
||||
|
||||
Reference in New Issue
Block a user