ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-04-23 16:09:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fuse sum_rows and div with topk-moe

Browse Source
This commit is contained in:
Kawrakow
2025-11-19 10:14:33 +00:00
parent 054c31cf8f
commit f514891418
3 changed files with 45 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/llama-build-context.cpp
View File

@@ -865,10 +865,6 @@ llm_expert_gating_func_type gating_op,
cb(weights, "ffn_moe_weights_softmax", il);
}
if (graph) {
ggml_build_forward_expand(graph, weights);
}
if (norm_w) {
weights = ggml_reshape_2d(ctx, weights, n_expert_used, n_tokens);
@@ -890,6 +886,10 @@ llm_expert_gating_func_type gating_op,
cb(weights, "ffn_moe_weights_scaled", il);
}
if (graph) {
ggml_build_forward_expand(graph, weights);
}
cur = ggml_reshape_3d(ctx, cur, n_embd, 1, n_tokens);
if (weight_before_ffn) {