8 lines
203 B
Python
8 lines
203 B
Python
import torch
|
|
|
|
|
|
def lightning_attention_decode(q, k, v, past_kv, slope, output, new_kv):
|
|
torch.ops.sgl_kernel.lightning_attention_decode.default(
|
|
q, k, v, past_kv, slope, output, new_kv
|
|
)
|