Skip to content

[omni-vlm example] reset model in every inerence step to avoid nosense output. #18

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 1 commit into from
Nov 11, 2024
Merged

[omni-vlm example] reset model in every inerence step to avoid nosense output. #18

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
10 changes: 7 additions & 3 deletions examples/omni-vlm/omni-vlm-cli.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,10 @@
#include <cstdlib>
#include <cstring>
#include <vector>
// #include <iostream>
//
// using std::cout;
// using std::endl;

static bool eval_tokens(struct llama_context * ctx_llama, std::vector<llama_token> tokens, int n_batch, int * n_past) {
int N = (int) tokens.size();
Expand Down Expand Up @@ -283,9 +287,9 @@ int main(int argc, char ** argv) {
return 1;
}

auto * ctx_omnivlm = omnivlm_init_context(&params, model);

for (auto & image : params.image) {
auto * ctx_omnivlm = omnivlm_init_context(&params, model);
auto * image_embed = load_image(ctx_omnivlm, &params, image);
if (!image_embed) {
LOG_TEE("%s: failed to load image %s. Terminating\n\n", __func__, image.c_str());
Expand All @@ -296,9 +300,9 @@ int main(int argc, char ** argv) {

llama_print_timings(ctx_omnivlm->ctx_llama);
omnivlm_image_embed_free(image_embed);
ctx_omnivlm->model = NULL;
omnivlm_free(ctx_omnivlm);
}
ctx_omnivlm->model = NULL;
omnivlm_free(ctx_omnivlm);

llama_free_model(model);

Expand Down
6 changes: 5 additions & 1 deletion examples/omni-vlm/omni-vlm-wrapper.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -244,10 +244,11 @@ void omnivlm_init(const char* llm_model_path, const char* projector_model_path,
fprintf(stderr, "%s: error: failed to init omnivlm model\n", __func__);
throw std::runtime_error("Failed to init omnivlm model");
}
ctx_omnivlm = omnivlm_init_context(&params, model);
}

const char* omnivlm_inference(const char *prompt, const char *imag_path) {
ctx_omnivlm = omnivlm_init_context(&params, model);

std::string image = imag_path;
params.prompt = prompt;

Expand All @@ -270,6 +271,9 @@ const char* omnivlm_inference(const char *prompt, const char *imag_path) {

// llama_perf_print(ctx_omnivlm->ctx_llama, LLAMA_PERF_TYPE_CONTEXT);
omnivlm_image_embed_free(image_embed);
ctx_omnivlm->model = nullptr;
omnivlm_free(ctx_omnivlm);
ctx_omnivlm = nullptr;

return ret_chars;
}
Expand Down