Commit a2b301831 for llama.cpp

commit a2b301831d05b07860de508d229d430207127802
Author: bosh <98094229+boshjerns@users.noreply.github.com>
Date:   Sun Nov 8 13:11:07 2026 +0700

    speculative : do not apply the MoE cache to the draft model (#31497)

diff --git a/examples/speculative/speculative.cpp b/examples/speculative/speculative.cpp
index 8f347b0d4..f3c6bec6f 100644
--- a/examples/speculative/speculative.cpp
+++ b/examples/speculative/speculative.cpp
@@ -103,6 +103,9 @@ int main(int argc, char ** argv) {
     params.cpuparams_batch.n_threads = params.speculative.draft.cpuparams_batch.n_threads;
     params.tensor_buft_overrides     = params.speculative.draft.tensor_buft_overrides;

+    // the MoE cache is only used by the target context
+    params.moe_cache_size = 0;
+
     auto llama_init_dft = common_init_from_params(params);

     model_dft = llama_init_dft->model();