diff --git a/ggml/src/gguf.cpp b/ggml/src/gguf.cpp
index 5e1986182515..61026df1c6ab 100644
--- a/ggml/src/gguf.cpp
+++ b/ggml/src/gguf.cpp
@@ -15,6 +15,8 @@
 #include <string>
 #include <vector>
 
+#include "moz-overrides.h"
+
 #define GGUF_MAX_STRING_LENGTH  (1024*1024*1024)
 #define GGUF_MAX_ARRAY_ELEMENTS (1024*1024*1024)

diff --git a/src/llama-adapter.cpp b/src/llama-adapter.cpp
index 3e0fe66afff7..95f50e05ccd8 100644
--- a/src/llama-adapter.cpp
+++ b/src/llama-adapter.cpp
@@ -9,6 +9,8 @@
 #include <sstream>
 #include <stdexcept>
 
+#include "moz-overrides.h"
+
 // vec
 
 ggml_tensor * llama_adapter_cvec::tensor_for(int il) const {
diff --git a/src/llama-chat.cpp b/src/llama-chat.cpp
index 6d822ec62d6b..c83e6b5ba0d4 100644
--- a/src/llama-chat.cpp
+++ b/src/llama-chat.cpp
@@ -6,6 +6,8 @@
 #include <sstream>
 #include <algorithm>
 
+#include "moz-overrides.h"
+
 #if __cplusplus >= 202000L
     #define LU8(x) (const char*)(u8##x)
 #else
diff --git a/src/llama-context.cpp b/src/llama-context.cpp
index 168dbabd7667..2de0c7ad45f4 100644
--- a/src/llama-context.cpp
+++ b/src/llama-context.cpp
@@ -18,6 +18,8 @@
 #include <limits>
 #include <stdexcept>
 
+#include "moz-overrides.h"
+
 //
 // llama_context
 //
diff --git a/src/llama-grammar.cpp b/src/llama-grammar.cpp
index badcbfd0fbb6..3872a7641f70 100644
--- a/src/llama-grammar.cpp
+++ b/src/llama-grammar.cpp
@@ -10,7 +10,10 @@
 #include <set>
 #include <stdexcept>
 
+#include "moz-overrides.h"
+
 #define MAX_REPETITION_THRESHOLD 2000
+
 //
 // helpers
 //
diff --git a/src/llama-kv-cache.cpp b/src/llama-kv-cache.cpp
index 2802103bdd82..289a6a0c086e 100644
--- a/src/llama-kv-cache.cpp
+++ b/src/llama-kv-cache.cpp
@@ -13,6 +13,8 @@
 #include <map>
 #include <stdexcept>
 
+#include "moz-overrides.h"
+
 static bool ggml_is_power_of_2(int n) {
     return (n & (n - 1)) == 0;
 }
diff --git a/src/llama-memory-recurrent.cpp b/src/llama-memory-recurrent.cpp
index 6a4892fb471d..f300c1afacbb 100644
--- a/src/llama-memory-recurrent.cpp
+++ b/src/llama-memory-recurrent.cpp
@@ -13,6 +13,8 @@
 #include <map>
 #include <stdexcept>
 
+#include "moz-overrides.h"
+
 //
 // llama_memory_recurrent
 //
diff --git a/src/llama-mmap.cpp b/src/llama-mmap.cpp
index ed572da7fb54..0e4c620c5dd4 100644
--- a/src/llama-mmap.cpp
+++ b/src/llama-mmap.cpp
@@ -40,6 +40,8 @@
 #include <TargetConditionals.h>
 #endif
 
+#include "moz-overrides.h"
+
 #ifdef _WIN32
 #    define llama_mmap_ftell _ftelli64
 #    define llama_mmap_fseek _fseeki64
diff --git a/src/llama-model-loader.cpp b/src/llama-model-loader.cpp
index 474cabdfc095..4456259be12d 100644
--- a/src/llama-model-loader.cpp
+++ b/src/llama-model-loader.cpp
@@ -13,6 +13,8 @@
 #include <future>
 #include <regex>
 
+#include "moz-overrides.h"
+
 static const size_t kiB = 1024;
 static const size_t MiB = 1024*kiB;
 static const size_t GiB = 1024*MiB;
diff --git a/src/llama-model-loader.h b/src/llama-model-loader.h
index c476026d3e51..1a1c4c4db4db 100644
--- a/src/llama-model-loader.h
+++ b/src/llama-model-loader.h
@@ -39,12 +39,14 @@ struct llama_model_loader {
         llama_tensor_weight(const llama_file * file, uint16_t idx, const struct gguf_context * gguf_ctx, ggml_tensor * tensor) : idx(idx), tensor(tensor) {
             const int tensor_idx = gguf_find_tensor(gguf_ctx,  ggml_get_name(tensor));
             if (tensor_idx < 0) {
-                throw std::runtime_error(format("tensor '%s' not found in the model", ggml_get_name(tensor)));
+                // throw std::runtime_error(format("tensor '%s' not found in the model", ggml_get_name(tensor)));
+              std::abort();
             }
 
             offs = gguf_get_data_offset(gguf_ctx) + gguf_get_tensor_offset(gguf_ctx, tensor_idx);
             if (offs + ggml_nbytes(tensor) < offs || offs + ggml_nbytes(tensor) > file->size()) {
-                throw std::runtime_error(format("tensor '%s' data is not within the file bounds, model is corrupted or incomplete", ggml_get_name(tensor)));
+              //throw std::runtime_error(format("tensor '%s' data is not within the file bounds, model is corrupted or incomplete", ggml_get_name(tensor)));
+              std::abort();
             }
         }
     };
diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index c52875533905..f9a1a7a446b1 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -683,6 +683,8 @@ struct ggml_backend_meta_split_state llama_meta_device_get_split_state(const str
     GGML_UNUSED(userdata);
 }
 
+#include "moz-overrides.h"
+
 const char * llm_type_name(llm_type type) {
     switch (type) {
         case LLM_TYPE_14M:           return "14M";
diff --git a/src/llama-sampler.cpp b/src/llama-sampler.cpp
index 9bbc5dbde247..82f2babcab9b 100644
--- a/src/llama-sampler.cpp
+++ b/src/llama-sampler.cpp
@@ -20,6 +20,8 @@
 #include <unordered_map>
 #include <stdexcept>
 
+#include "moz-overrides.h"
+
 // the ring buffer works similarly to std::deque, but with a fixed capacity
 template<typename T>
 struct ring_buffer {
diff --git a/src/llama-vocab.cpp b/src/llama-vocab.cpp
index 6e78a3f6c0ea..924957674a26 100644
--- a/src/llama-vocab.cpp
+++ b/src/llama-vocab.cpp
@@ -21,6 +21,8 @@
 #include <set>
 #include <unordered_map>
 
+#include "moz-overrides.h"
+
 //
 // helpers
 //
diff --git a/src/llama.cpp b/src/llama.cpp
index a67fa8039a48..2b5726eb0f73 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -26,6 +26,8 @@
 #include <stdexcept>
 #include <vector>
 
+#include "moz-overrides.h"
+
 #if defined(_MSC_VER)
 #pragma warning(disable: 4244 4267) // possible loss of data
 #endif
diff --git a/src/unicode.cpp b/src/unicode.cpp
index b02ecdc930fa..f7870412a3f4 100644
--- a/src/unicode.cpp
+++ b/src/unicode.cpp
@@ -13,6 +13,8 @@
 #include <utility>
 #include <vector>
 
+#include "moz-overrides.h"
+
 size_t unicode_len_utf8(char src) {
     const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
     uint8_t highbits = static_cast<uint8_t>(src) >> 4;
diff --git a/src/models/models.h b/src/models/models.h
index 19a4d3c5eaf4..2ac8415a3639 100644
--- a/src/models/models.h
+++ b/src/models/models.h
@@ -7,8 +7,6 @@
 // note: almost all graphs require at least sqrtf, so include cmath globally
 #include <cmath>
 
-#include "moz-overrides.h"
-
 //
 // base classes
 //