diff --git a/common/log.h b/common/log.h
index 6934c57b2225c..af54bb347afc3 100644
--- a/common/log.h
+++ b/common/log.h
@@ -596,7 +596,7 @@ inline std::string log_var_to_string_impl(bool var)
     return var ? "true" : "false";
 }
 
-inline std::string log_var_to_string_impl(std::string var)
+inline std::string log_var_to_string_impl(const std::string & var)
 {
     return var;
 }
diff --git a/common/ngram-cache.cpp b/common/ngram-cache.cpp
index 3ca112ef1613d..9ad4a243265b0 100644
--- a/common/ngram-cache.cpp
+++ b/common/ngram-cache.cpp
@@ -59,7 +59,7 @@ constexpr int draft_min_sample_size_strict[LLAMA_NGRAM_MAX] = { 4,  3,  2,  2};
 constexpr int     draft_min_percent_strict[LLAMA_NGRAM_MAX] = {75, 66, 66, 66};
 
 // Helper function that tries to draft a token from only the static ngram cache:
-static llama_token try_draft(llama_ngram_cache & nc_static, const llama_ngram ngram_static) {
+static llama_token try_draft(llama_ngram_cache & nc_static, const llama_ngram & ngram_static) {
     llama_ngram_cache::iterator part_static_it = nc_static.find(ngram_static);
     if (part_static_it == nc_static.end()) {
         return -1;
diff --git a/ggml.c b/ggml.c
index b96a82a41517d..1471369a4235f 100644
--- a/ggml.c
+++ b/ggml.c
@@ -3009,7 +3009,7 @@ size_t ggml_used_mem(const struct ggml_context * ctx) {
     return ctx->objects_end == NULL ? 0 : ctx->objects_end->offs + ctx->objects_end->size;
 }
 
-size_t ggml_set_scratch(struct ggml_context * ctx, struct ggml_scratch scratch) {
+size_t ggml_set_scratch(struct ggml_context * ctx, const struct ggml_scratch scratch) {
     const size_t result = ctx->scratch.data ? ctx->scratch.offs : 0;
 
     ctx->scratch = scratch;
@@ -20646,7 +20646,7 @@ static void ggml_opt_acc_grad(int np, struct ggml_tensor * const ps[], float * g
 static enum ggml_opt_result ggml_opt_adam(
         struct ggml_context * ctx,
         struct ggml_opt_context * opt,
-        struct ggml_opt_params params,
+        const struct ggml_opt_params params,
         struct ggml_tensor * f,
         struct ggml_cgraph * gf,
         struct ggml_cgraph * gb,
@@ -21345,7 +21345,7 @@ struct ggml_opt_params ggml_opt_default_params(enum ggml_opt_type type) {
 GGML_API void ggml_opt_init(
         struct ggml_context * ctx,
         struct ggml_opt_context * opt,
-        struct ggml_opt_params params,
+        const struct ggml_opt_params params,
         int64_t nx) {
     opt->ctx = ctx;
     opt->params = params;
@@ -21417,7 +21417,7 @@ GGML_API void ggml_opt_init(
 
 enum ggml_opt_result ggml_opt(
         struct ggml_context * ctx,
-        struct ggml_opt_params params,
+        const struct ggml_opt_params params,
         struct ggml_tensor * f) {
     bool free_ctx = false;
     if (ctx == NULL) {