ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

mtmd : refactor llava-uhd preprocessing logic (#14247)

Browse source 
* mtmd : refactor llava-uhd preprocessing logic

* fix editorconfig
This commit is contained in:
Xuan-Son Nguyen 2025-06-18 10:43:57 +02:00 • committed by GitHub
parent 95402553a5
commit 413977de32
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 110 additions and 80 deletions
Download patch file Download diff file Expand all files Collapse all files

6
tools/mtmd/mtmd.cpp
View file

@ -501,7 +501,10 @@ struct mtmd_tokenizer {
|| ctx->slice_tmpl == MTMD_SLICE_TMPL_MINICPMV_2_6
|| ctx->slice_tmpl == MTMD_SLICE_TMPL_LLAMA4
) {
const int n_col = batch_f32.grid_x;
const int n_row = batch_f32.grid_y;
// split batch into chunks of single images
// NOTE: batch_f32 will be invalidated after this call
auto chunks = split_batch_to_chunk(std::move(batch_f32), bitmap->id);
GGML_ASSERT(chunks.size() > 0);
@ -521,8 +524,7 @@ struct mtmd_tokenizer {
// add slices (or tiles)
if (!chunks.empty()) {
const int n_col = batch_f32.grid_x;
const int n_row = batch_f32.grid_y;
GGML_ASSERT((int)chunks.size() == n_row * n_col);
if (ctx->tok_slices_start != LLAMA_TOKEN_NULL) {
add_text({ctx->tok_slices_start});
}