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

sync : ggml (#5452)

Browse source 
* ggml-alloc : v3 (ggml/727)

* ggml-alloc v3

ggml-ci

* fix ci

ggml-ci

* whisper : check for backend buffer allocation failures

* whisper : avoid leaks when initialization fails

* cleanup

ggml-ci

* style fixes

ggml-ci

* sync : ggml

* update llama.cpp, clip.cpp, export-lora.cpp

* update finetune.cpp, train-text-from-scratch.cpp

ggml-ci

* ggml-backend : reduce alignment to 32 to match gguf and fix mmap

---------

Co-authored-by: slaren <slarengh@gmail.com>
This commit is contained in:
Georgi Gerganov 2024-02-12 09:16:06 +02:00 • committed by GitHub
parent 3bdc4cd0f5
commit 3b169441df
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 1287 additions and 1362 deletions
Download patch file Download diff file Expand all files Collapse all files

19
examples/export-lora/export-lora.cpp
View file

@ -337,24 +337,14 @@ static bool apply_lora(struct ggml_tensor * tensor, struct lora_data * lora, int
params.mem_buffer = NULL;
params.no_alloc = true;
struct ggml_context * ctx = NULL;
struct ggml_allocr * alloc = NULL;
struct ggml_cgraph * gf = NULL;
struct ggml_gallocr * alloc = NULL;
struct ggml_cgraph * gf = NULL;
ctx = ggml_init(params);
alloc = ggml_allocr_new_measure(tensor_alignment);
alloc = ggml_gallocr_new(ggml_backend_cpu_buffer_type());
gf = build_graph_lora(ctx, tensor, lora_a, lora_b, scaling);
size_t alloc_size = ggml_allocr_alloc_graph(alloc, gf);
ggml_allocr_free(alloc);
ggml_free(ctx);
static std::vector<uint8_t> data_compute;
data_compute.resize(alloc_size + tensor_alignment);
ctx = ggml_init(params);
alloc = ggml_allocr_new(data_compute.data(), data_compute.size(), tensor_alignment);
gf = build_graph_lora(ctx, tensor, lora_a, lora_b, scaling);
ggml_allocr_alloc_graph(alloc, gf);
ggml_allocr_free(alloc);
ggml_gallocr_alloc_graph(alloc, gf);
struct ggml_cplan cplan = ggml_graph_plan(gf, n_threads);
static std::vector<uint8_t> data_work;
@ -363,6 +353,7 @@ static bool apply_lora(struct ggml_tensor * tensor, struct lora_data * lora, int
ggml_graph_compute(gf, &cplan);
ggml_gallocr_free(alloc);
ggml_free(ctx);
return true;
}