[Feature] Decode Context Parallel support for GPU model runner v2 (#34179)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2026-02-17 16:27:15 -08:00
committed by GitHub
parent be3af2d29e
commit ab33d2a629
5 changed files with 117 additions and 3 deletions
+4
View File
@@ -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