[GPTOSS][DP/EP][Marlin] Enable GPTOSS DP/EP using Marlin kernels (#25488)

Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Varun Sundar Rabindranath
2025-10-03 20:13:13 -04:00
committed by GitHub
parent 767cbb011d
commit 7ef40bb983
9 changed files with 264 additions and 101 deletions

View File

@@ -1780,7 +1780,7 @@ class TritonExperts(mk.FusedMoEPermuteExpertsUnpermute):
torch.float32, torch.float16, torch.bfloat16, torch.float8_e4m3fn
]
E, num_tokens, N, K, top_k_num = mk._moe_problem_size(
E, num_tokens, N, K, top_k_num = self.moe_problem_size(
hidden_states, w1, w2, topk_ids)
if global_num_experts == -1: