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

ggml : hide ggml_object, ggml_cgraph, ggml_hash_set (#9408)

Browse source 
* ggml : hide ggml_object, ggml_cgraph, ggml_hash_set

ggml-ci

* ggml : add ggml-impl.h to backends

* ggml : fix compiler warnings

ggml-ci

* ggml : add assert upon adding nodes
This commit is contained in:
Georgi Gerganov 2024-09-12 14:23:49 +03:00 • committed by GitHub
parent c9c8575a1a
commit d6a04f872d
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
18 changed files with 170 additions and 129 deletions
Download patch file Download diff file Expand all files Collapse all files

2
examples/llava/llava.cpp
View file

@ -184,7 +184,7 @@ static bool clip_llava_handle_patches(clip_ctx * ctx_clip, std::vector<float *>
// ggml_tensor_printf(flatten,"flatten",__LINE__,false,false);
ggml_build_forward_expand(gf, flatten);
ggml_graph_compute_with_ctx(model.ctx, gf, 1);
struct ggml_tensor* result = gf->nodes[gf->n_nodes - 1];
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
// append without newline tokens (default behavior in llava_arch when not using unpad ):