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

llava/mtmd : fixes to fully support dl backends (#13303)

Browse source
This commit is contained in:
Diego Devesa 2025-05-04 17:05:20 +02:00 • committed by GitHub
parent 9f2da5871f
commit 86bd60d3fe
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 15 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
tools/llava/llava.cpp
View file

@ -2,6 +2,7 @@
#include "llava.h"
#include "llama.h"
#include "ggml-cpp.h"
#include <algorithm>
#include <cerrno>
@ -209,7 +210,10 @@ static bool clip_llava_handle_patches(clip_ctx * ctx_clip, std::vector<float *>
struct ggml_tensor *flatten = ggml_view_2d(model.ctx, permuted_cont, clip_n_mmproj_embd(ctx_clip), num_patches_height * num_patches_width * num_patches_per_side * num_patches_per_side, size_ele * clip_n_mmproj_embd(ctx_clip), 0);
// ggml_tensor_printf(flatten,"flatten",__LINE__,false,false);
ggml_build_forward_expand(gf, flatten);
ggml_graph_compute_with_ctx(model.ctx, gf, 1);
ggml_backend_ptr backend { ggml_backend_init_by_type(GGML_BACKEND_DEVICE_TYPE_CPU, nullptr) };
ggml_backend_graph_compute(backend.get(), gf);
struct ggml_tensor* result = ggml_graph_node(gf, -1);
memcpy(image_embd_out, image_embd_v[0], clip_embd_nbytes(ctx_clip)); // main image as global context