vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Added GPU support on qwen2vl readme

Browse source 
Undo changes on qwen2vl-cli
This commit is contained in:
sami 2025-02-09 15:41:11 +07:00
parent 8777473a43
commit 185e1b107e
2 changed files with 4 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

5
examples/llava/README-qwen2vl.md
View file

@ -57,6 +57,7 @@ Now the model is ready to use in the `model_path` directory. You can quantize th
*Have fun with the models ! :)* *Have fun with the models ! :)*
## Limitations ## Current limitations
* Currently, only support the image to be in the very beginning of the input prompt to the LLM. * This only supports the image to be in the very beginning of the input prompt to the LLM.
* The vision model (clip.cpp)'s GPU backend support, which Qwen2VL uses, is disabled.

11
examples/llava/qwen2vl-cli.cpp
View file

@ -524,7 +524,7 @@ int main(int argc, char ** argv) {
common_init(); common_init();
if (params.mmproj.empty()) { if (params.mmproj.empty() || (params.image.empty() && !prompt_contains_image(params.prompt))) {
print_usage(argc, argv); print_usage(argc, argv);
return 1; return 1;
} }
@ -547,15 +547,6 @@ int main(int argc, char ** argv) {
llava_image_embed_free(image_embed); llava_image_embed_free(image_embed);
ctx_llava->model = NULL; ctx_llava->model = NULL;
llava_free(ctx_llava); llava_free(ctx_llava);
} else if (params.image.empty()) {
auto ctx_llava = llava_init_context(&params, model);
// process the prompt
process_prompt(ctx_llava, nullptr, &params, params.prompt);
llama_perf_context_print(ctx_llava->ctx_llama);
ctx_llava->model = NULL;
llava_free(ctx_llava);
#ifndef NDEBUG #ifndef NDEBUG
} else if (params.image[0].empty()) { } else if (params.image[0].empty()) {
auto ctx_llava = llava_init_context(&params, model); auto ctx_llava = llava_init_context(&params, model);