Skip to content

llama-mtmd-cli: Sigint rework in mtmd vision example #13080

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 5 commits into from
Apr 23, 2025
Merged

llama-mtmd-cli: Sigint rework in mtmd vision example #13080

Changes from 2 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
cf83094
Sigint rework in mtmd vision example
Apr 23, 2025
160b170
Applied suggestions on mtmd-cli PR
Apr 23, 2025
cadaa28
Forgot to invert one of the conditions
pl752 Apr 23, 2025
f6951e7
Update examples/llava/mtmd-cli.cpp
ngxson Apr 23, 2025
0e58d40
Removed redundant exit check
pl752 Apr 23, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 25 additions & 7 deletions examples/llava/mtmd-cli.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,9 @@
#include <signal.h>
#endif

static bool g_is_generating = false;
//volatile, because of signal being an interrupt
static volatile bool g_is_generating = false;
static volatile bool g_is_interrupted = false;

/**
* Please note that this is NOT a production-ready stuff.
Expand All @@ -50,8 +52,10 @@ static void sigint_handler(int signo) {
g_is_generating = false;
} else {
console::cleanup();
LOG("\nInterrupted by user\n");
_exit(130);
if (g_is_interrupted) {
_exit(1);
}
g_is_interrupted = true;
}
}
}
Expand Down Expand Up @@ -167,7 +171,7 @@ struct decode_embd_batch {
static int generate_response(mtmd_cli_context & ctx, common_sampler * smpl, int n_predict) {
llama_tokens generated_tokens;
for (int i = 0; i < n_predict; i++) {
if (i > n_predict || !g_is_generating) {
if (i > n_predict || !g_is_generating || g_is_interrupted) {
printf("\n");
break;
}
Expand All @@ -184,6 +188,11 @@ static int generate_response(mtmd_cli_context & ctx, common_sampler * smpl, int
printf("%s", common_token_to_piece(ctx.lctx, token_id).c_str());
fflush(stdout);

if (g_is_interrupted) {
printf("\n");
break;
}

// eval the token
common_batch_clear(ctx.batch);
common_batch_add(ctx.batch, token_id, ctx.n_past++, {0}, true);
Expand All @@ -206,6 +215,7 @@ static int eval_message(mtmd_cli_context & ctx, common_chat_msg & msg, std::vect
LOG_DBG("formatted_chat.prompt: %s\n", formatted_chat.prompt.c_str());

for (auto & fname : images_fname) {
if (g_is_interrupted) return 0;
mtmd_bitmap bitmap;
if (mtmd_helper_bitmap_init_from_file(fname.c_str(), bitmap)) {
LOG_ERR("Unable to load image %s\n", fname.c_str());
Expand All @@ -219,6 +229,9 @@ static int eval_message(mtmd_cli_context & ctx, common_chat_msg & msg, std::vect
text.add_special = add_bos;
text.parse_special = true;
mtmd_input_chunks chunks;

if (g_is_interrupted) return 0;

int32_t res = mtmd_tokenize(ctx.ctx_vision.get(), chunks, text, bitmaps);
if (res != 0) {
LOG_ERR("Unable to tokenize prompt, res = %d\n", res);
Expand Down Expand Up @@ -276,6 +289,8 @@ int main(int argc, char ** argv) {
#endif
}

if (g_is_interrupted) return 130;

if (is_single_turn) {
g_is_generating = true;
if (params.prompt.find("<__image__>") == std::string::npos) {
Expand All @@ -287,7 +302,7 @@ int main(int argc, char ** argv) {
if (eval_message(ctx, msg, params.image, true)) {
return 1;
}
if (generate_response(ctx, smpl, n_predict)) {
if (g_is_interrupted && generate_response(ctx, smpl, n_predict)) {
return 1;
}

Expand All @@ -302,12 +317,13 @@ int main(int argc, char ** argv) {
std::vector<std::string> images_fname;
std::string content;

while (true) {
while (!g_is_interrupted) {
g_is_generating = false;
LOG("\n> ");
console::set_display(console::user_input);
std::string line;
console::readline(line, false);
if (g_is_interrupted) break;
console::set_display(console::reset);
line = string_strip(line);
if (line.empty()) {
Expand Down Expand Up @@ -335,6 +351,7 @@ int main(int argc, char ** argv) {
msg.role = "user";
msg.content = content;
int ret = eval_message(ctx, msg, images_fname, is_first_msg);
if (g_is_interrupted) break;
if (ret == 2) {
// non-fatal error
images_fname.clear();
Expand All @@ -352,6 +369,7 @@ int main(int argc, char ** argv) {
is_first_msg = false;
}
}
if (g_is_interrupted) LOG("\nInterrupted by user\n");
llama_perf_context_print(ctx.lctx);
return 0;
return g_is_interrupted ? 130 : 0;
}