Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion custom_ops/gpu_ops/noaux_tc_redundant.cu
Original file line number Diff line number Diff line change
Expand Up @@ -61,7 +61,8 @@ std::vector<paddle::Tensor> NoauxTcRedundant(
redundant_ep_rank_num_plus_one,
stream);

return {scores, topk_values, topk_indices};
// Return 4 values to match PD_BUILD_STATIC_OP Outputs definition
return {scores, topk_values, topk_indices, tokens_per_expert_stats_list};
}

std::vector<paddle::DataType> NoauxTcRedundantInferDtype(
Expand Down
4 changes: 3 additions & 1 deletion fastdeploy/model_executor/layers/moe/moe.py
Original file line number Diff line number Diff line change
Expand Up @@ -104,7 +104,9 @@ def get_moe_scores(
routed_scaling_factor,
)
else:
scores, topk_values, topk_idx = noaux_tc_redundant(
# noaux_tc_redundant returns 4 values: scores, topk_values, topk_idx,
# and tokens_per_expert_stats_list_out (inplace updated)
scores, topk_values, topk_idx, _ = noaux_tc_redundant(
scores,
scores_with_bias,
expert_id_to_ep_rank_array,
Expand Down
Loading