Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

tests : fix log verbosity for test-llama-archs (#28147)

Browse Source 
* tests : fix log verbosity for test-llama-archs

* cont : naming

* cont : add note
This commit is contained in:
Georgi Gerganov
2026-09-01 13:07:12 +03:00
committed by GitHub
parent 1b89a43e38
commit d086dbb348
3 changed files with 47 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files
common/log.cpp
+2 -2
View File
@@ -438,7 +438,7 @@ void common_log_flush(struct common_log * log) {
log->resume(); log->resume();
} }
static int common_get_verbosity(enum ggml_log_level level) { int common_log_get_verbosity(enum ggml_log_level level) {
switch (level) { switch (level) {
case GGML_LOG_LEVEL_DEBUG: return LOG_LEVEL_DEBUG; case GGML_LOG_LEVEL_DEBUG: return LOG_LEVEL_DEBUG;
case GGML_LOG_LEVEL_INFO: return LOG_LEVEL_TRACE; case GGML_LOG_LEVEL_INFO: return LOG_LEVEL_TRACE;
@@ -452,7 +452,7 @@ static int common_get_verbosity(enum ggml_log_level level) {
} }
void common_log_default_callback(enum ggml_log_level level, const char * text, void * /*user_data*/) { void common_log_default_callback(enum ggml_log_level level, const char * text, void * /*user_data*/) {
auto verbosity = common_get_verbosity(level); auto verbosity = common_log_get_verbosity(level);
if (verbosity <= common_log_verbosity_thold) { if (verbosity <= common_log_verbosity_thold) {
common_log_add(common_log_main(), level, "%s", text); common_log_add(common_log_main(), level, "%s", text);
} }
common/log.h
+2
View File
@@ -43,6 +43,8 @@ int common_log_get_verbosity_thold(void);
void common_log_set_verbosity_thold(int verbosity); // not thread-safe void common_log_set_verbosity_thold(int verbosity); // not thread-safe
int common_log_get_verbosity(enum ggml_log_level level);
void common_log_default_callback(enum ggml_log_level level, const char * text, void * user_data); void common_log_default_callback(enum ggml_log_level level, const char * text, void * user_data);
// the common_log uses an internal worker thread to print/write log messages // the common_log uses an internal worker thread to print/write log messages
tests/test-llama-archs.cpp
+43 -26
View File
@@ -65,7 +65,7 @@ static void set_tensor_data(struct ggml_tensor * tensor, void * userdata) {
} }
static void usage(char ** argv) { static void usage(char ** argv) {
printf("Usage: %s [-a/--arch arch] [-s/--seed seed] [-o/--out dir] [-v/--verbose] [-h/--help]\n", argv[0]); printf("Usage: %s [-a/--arch arch] [-s/--seed seed] [-o/--out dir] [-v N] [-h/--help]\n", argv[0]);
} }
static std::vector<llama_token> get_tokens(const uint32_t n_tokens, const uint32_t n_vocab, const size_t seed){ static std::vector<llama_token> get_tokens(const uint32_t n_tokens, const uint32_t n_vocab, const size_t seed){
@@ -535,22 +535,27 @@ static bool arch_supported(const llm_arch arch) {
return true; return true;
} }
static int save_models(const llm_arch target_arch, const size_t seed, const ggml_log_level log_level, const std::string & dir) { static int save_models(const llm_arch target_arch, const size_t seed, const int verbosity, const std::string & dir) {
struct user_data_t { struct user_data_t {
struct { struct {
ggml_log_callback callback; ggml_log_callback callback;
void * user_data; void * user_data;
} original_logger; } log_old;
ggml_log_level min_level; // prints below this log level go to debug log
int verbosity;
user_data_t(int verbosity) : verbosity(verbosity) {
llama_log_get(&log_old.callback, &log_old.user_data);
}
}; };
user_data_t ud; user_data_t ud(verbosity);
llama_log_get(&ud.original_logger.callback, &ud.original_logger.user_data);
ud.min_level = log_level;
llama_log_set([](ggml_log_level level, const char * text, void * user_data) { llama_log_set([](ggml_log_level level, const char * text, void * user_data) {
const user_data_t * ud = (const user_data_t *) user_data; const user_data_t * ud = (const user_data_t *) user_data;
const ggml_log_level level_eff = level >= ud->min_level ? level : GGML_LOG_LEVEL_DEBUG; int verbosity = common_log_get_verbosity(level);
ud->original_logger.callback(level_eff, text, ud->original_logger.user_data); if (verbosity <= ud->verbosity) {
ud->log_old.callback(level, text, ud->log_old.user_data);
}
}, &ud); }, &ud);
for (const llm_arch & arch : llm_arch_all()) { for (const llm_arch & arch : llm_arch_all()) {
@@ -584,26 +589,31 @@ static int save_models(const llm_arch target_arch, const size_t seed, const ggml
llama_model_save_to_file(model_and_ctx.first.get(), path.c_str()); llama_model_save_to_file(model_and_ctx.first.get(), path.c_str());
} }
} }
llama_log_set(ud.original_logger.callback, ud.original_logger.user_data); llama_log_set(ud.log_old.callback, ud.log_old.user_data);
return 0; return 0;
} }
static int test_backends(const llm_arch target_arch, const size_t seed, const ggml_log_level log_level) { static int test_backends(const llm_arch target_arch, const size_t seed, const int verbosity) {
struct user_data_t { struct user_data_t {
struct { struct {
ggml_log_callback callback; ggml_log_callback callback;
void * user_data; void * user_data;
} original_logger; } log_old;
ggml_log_level min_level; // prints below this log level go to debug log
int verbosity;
user_data_t(int verbosity) : verbosity(verbosity) {
llama_log_get(&log_old.callback, &log_old.user_data);
}
}; };
user_data_t ud; user_data_t ud(verbosity);
llama_log_get(&ud.original_logger.callback, &ud.original_logger.user_data);
ud.min_level = log_level;
llama_log_set([](ggml_log_level level, const char * text, void * user_data) { llama_log_set([](ggml_log_level level, const char * text, void * user_data) {
const user_data_t * ud = (const user_data_t *) user_data; const user_data_t * ud = (const user_data_t *) user_data;
const ggml_log_level level_eff = level >= ud->min_level ? level : GGML_LOG_LEVEL_DEBUG; int verbosity = common_log_get_verbosity(level);
ud->original_logger.callback(level_eff, text, ud->original_logger.user_data); if (verbosity <= ud->verbosity) {
ud->log_old.callback(level, text, ud->log_old.user_data);
}
}, &ud); }, &ud);
const std::vector<llama_token> tokens = get_tokens(128, 128, seed); const std::vector<llama_token> tokens = get_tokens(128, 128, seed);
@@ -749,20 +759,23 @@ static int test_backends(const llm_arch target_arch, const size_t seed, const gg
} }
} }
} }
llama_log_set(ud.original_logger.callback, ud.original_logger.user_data); llama_log_set(ud.log_old.callback, ud.log_old.user_data);
return all_ok ? 0 : 1; return all_ok ? 0 : 1;
} }
int main(int argc, char ** argv) { int main(int argc, char ** argv) {
// FIXME these tests are disabled in the CI for macOS-latest-cmake-arm64 because they are segfaulting // init the logger at max verbosity. filter with a custom callback respecting the user-configure verbosity
common_log_set_verbosity_thold(LOG_LEVEL_DEBUG);
common_init(); common_init();
std::random_device rd; std::random_device rd;
llm_arch arch = LLM_ARCH_UNKNOWN; llm_arch arch = LLM_ARCH_UNKNOWN;
size_t seed = rd(); size_t seed = rd();
ggml_log_level log_level = GGML_LOG_LEVEL_ERROR;
std::string out; std::string out;
int verbosity = LOG_LEVEL_ERROR;
for (int i = 1; i < argc; i++) { for (int i = 1; i < argc; i++) {
if (strcmp(argv[i], "-h") == 0 || strcmp(argv[i], "--help") == 0) { if (strcmp(argv[i], "-h") == 0 || strcmp(argv[i], "--help") == 0) {
usage(argv); usage(argv);
@@ -789,9 +802,13 @@ int main(int argc, char ** argv) {
return 1; return 1;
} }
} }
if (strcmp(argv[i], "-v") == 0 || strcmp(argv[i], "--verbose") == 0) { if (strcmp(argv[i], "-v") == 0) {
log_level = GGML_LOG_LEVEL_INFO; if (i + 1 < argc) {
continue; verbosity = std::stoull(argv[++i]);
} else {
usage(argv);
return 1;
}
} }
if (strcmp(argv[i], "-o") == 0 || strcmp(argv[i], "--out") == 0) { if (strcmp(argv[i], "-o") == 0 || strcmp(argv[i], "--out") == 0) {
if (i + 1 < argc) { if (i + 1 < argc) {
@@ -806,9 +823,9 @@ int main(int argc, char ** argv) {
try { try {
if (!out.empty()) { if (!out.empty()) {
return save_models(arch, seed, log_level, out); return save_models(arch, seed, verbosity, out);
} }
return test_backends(arch, seed, log_level); return test_backends(arch, seed, verbosity);
} catch (const std::exception & err) { } catch (const std::exception & err) {
fprintf(stderr, "encountered runtime error: %s\n", err.what()); fprintf(stderr, "encountered runtime error: %s\n", err.what());
return -1; return -1;