[Feature] Decode Context Parallel support for GPU model runner v2 (#34179)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
@@ -27,6 +27,10 @@ class InputBuffers:
|
||||
max_num_reqs + 1, dtype=torch.int32, device=device
|
||||
)
|
||||
self.seq_lens = torch.zeros(max_num_reqs, dtype=torch.int32, device=device)
|
||||
# DCP: per-request local seq_lens buffer
|
||||
self.dcp_local_seq_lens = torch.zeros(
|
||||
max_num_reqs, dtype=torch.int32, device=device
|
||||
)
|
||||
|
||||
|
||||
@dataclass
|
||||
|
||||
Reference in New Issue
Block a user