common: refactor common/debug to move abort_on_nan into base_callback_data

Passing bool abort_on_nan as template parameter for common_debug_cb_eval is unnecessary and creates an issue with LTO.
It should just be a member of the base_callback_data instead.

common/debug.cpp

@@ -47,8 +47,7 @@ static float common_ggml_get_float_value(const uint8_t * data,
 
 #define INDENT "    "
 
-template <bool abort>
-void common_debug_print_tensor(uint8_t * data, ggml_type type, const int64_t * ne, const size_t * nb, int64_t n) {
+void common_debug_print_tensor(uint8_t * data, ggml_type type, const int64_t * ne, const size_t * nb, int64_t n, bool abort_on_nan) {
     GGML_ASSERT(n > 0);
     float sum = 0;
     for (int64_t i3 = 0; i3 < ne[3]; i3++) {
@@ -94,7 +93,7 @@ void common_debug_print_tensor(uint8_t * data, ggml_type type, const int64_t * n
         LOG(INDENT "sum = %f\n", sum);
     }
 
-    if constexpr (abort) {
+    if (abort_on_nan) {
         if (std::isnan(sum)) {
             LOG("encountered NaN - aborting\n");
             exit(0);
@@ -112,7 +111,7 @@ void common_debug_print_tensor(uint8_t * data, ggml_type type, const int64_t * n
  * @param user_data user data to pass at each call back
  * @return true to receive data or continue the graph, false otherwise
  */
-template <bool abort_on_nan> bool common_debug_cb_eval(struct ggml_tensor * t, bool ask, void * user_data) {
+bool common_debug_cb_eval(struct ggml_tensor * t, bool ask, void * user_data) {
     auto * cb_data = (base_callback_data *) user_data;
 
     const struct ggml_tensor * src0 = t->src[0];
@@ -154,14 +153,8 @@ template <bool abort_on_nan> bool common_debug_cb_eval(struct ggml_tensor * t, b
 
     if (!ggml_is_quantized(t->type) && matches_filter) {
         uint8_t * data = is_host ? (uint8_t *) t->data : cb_data->data.data();
-        common_debug_print_tensor<abort_on_nan>(data, t->type, t->ne, t->nb, 3);
+        common_debug_print_tensor(data, t->type, t->ne, t->nb, 3, cb_data->abort_on_nan);
     }
 
     return true;
 }
-
-// Explicit template instantiations
-template bool common_debug_cb_eval<false>(ggml_tensor *, bool, void *);
-template bool common_debug_cb_eval<true>(ggml_tensor *, bool, void *);
-template void common_debug_print_tensor<false>(uint8_t *, ggml_type, const int64_t *, const size_t *, int64_t);
-template void common_debug_print_tensor<true>(uint8_t *, ggml_type, const int64_t *, const size_t *, int64_t);

common/debug.h

@@ -12,23 +12,26 @@
 // ne   - the tensor dimensions array
 // nb   - the tensor strides array
 // n    - the number of rows/columns to fully print
-template <bool abort_on_nan> void common_debug_print_tensor(uint8_t * data, ggml_type type, const int64_t * ne, const size_t * nb, int64_t n);
+// aon  - abort if NaN is encountered
+void common_debug_print_tensor(uint8_t * data, ggml_type type, const int64_t * ne, const size_t * nb, int64_t n, bool aon = false);
 
 // Intended to use as callback for ggml_backend_sched_eval_callback
 // prints tensors that are processed in the computation graph
 // by default prints all tensors, but can be configured by creating a `base_callback_data` instance with
 // non-empty filter_patterns. See examples/debug.ccp for possible usage patterns
-// The template parameter determines whether an error should be thrown whenever a NaN is encountered
+// `base_callback_data` contains `abort_on_nan` flag that determines whether an error should be thrown whenever a NaN is encountered
 // in a tensor (useful for stopping debug sessions on first erroneous tensor)
 // The callback data will be passed as the third parameter (user_data)
-template <bool abort_on_nan> bool common_debug_cb_eval(struct ggml_tensor * t, bool ask, void * user_data);
+bool common_debug_cb_eval(struct ggml_tensor * t, bool ask, void * user_data);
+
 struct base_callback_data {
     std::vector<uint8_t>    data;
     std::vector<std::regex> tensor_filters;
+    bool                    abort_on_nan{false};
 
     base_callback_data() = default;
 
-    base_callback_data(common_params & params, const std::vector<std::string> & filter_patterns) {
+    base_callback_data(common_params & params, const std::vector<std::string> & filter_patterns, bool abort_on_nan = false) {
         for (const auto & pattern : filter_patterns) {
             try {
                 std::string anchored_pattern = "^" + pattern;
@@ -37,7 +40,9 @@ struct base_callback_data {
                 throw std::runtime_error("Invalid regex pattern '" + pattern + "': " + e.what());
             }
         }
-        params.cb_eval           = common_debug_cb_eval<false>;
+        this->abort_on_nan = abort_on_nan;
+
+        params.cb_eval           = common_debug_cb_eval;
         params.cb_eval_user_data = this;
     }
 };

examples/eval-callback/eval-callback.cpp

@@ -53,7 +53,7 @@ int main(int argc, char ** argv) {
 
     // pass the callback to the backend scheduler
     // it will be executed for each node during the graph computation
-    params.cb_eval = common_debug_cb_eval<false>;
+    params.cb_eval = common_debug_cb_eval;
     params.cb_eval_user_data = &cb_data;
     params.warmup = false;
 

tools/mtmd/debug/mtmd-debug.cpp

@@ -89,7 +89,7 @@ int main(int argc, char ** argv) {
         {
             // always enable debug callback
             mparams.cb_eval_user_data = &cb_data;
-            mparams.cb_eval = common_debug_cb_eval<false>;
+            mparams.cb_eval = common_debug_cb_eval;
         }
         ctx_mtmd.reset(mtmd_init_from_file(clip_path, model, mparams));
         if (!ctx_mtmd.get()) {

tools/mtmd/mtmd-cli.cpp

@@ -145,7 +145,7 @@ struct mtmd_cli_context {
         mparams.image_max_tokens = params.image_max_tokens;
         if (std::getenv("MTMD_DEBUG_GRAPH") != nullptr) {
             mparams.cb_eval_user_data = &cb_data;
-            mparams.cb_eval = common_debug_cb_eval<false>;
+            mparams.cb_eval = common_debug_cb_eval;
         }
         ctx_vision.reset(mtmd_init_from_file(clip_path, model, mparams));
         if (!ctx_vision.get()) {