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

Windows fixes (#31)

Browse source 
* Apply fixes suggested to build on windows

Issue: https://github.com/ggerganov/llama.cpp/issues/22

* Remove unsupported VLAs

* MSVC: Remove features that are only available on MSVC C++20.

* Fix zero initialization of the other fields.

* Change the use of vector for stack allocations.
This commit is contained in:
Sebastián A 2023-03-12 17:15:00 -03:00 • committed by GitHub
parent 7027a97837
commit eb062bb012
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 30 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

12
main.cpp
View file

@ -209,8 +209,8 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
// create the ggml context
{
struct ggml_init_params params = {
.mem_size = ctx_size,
.mem_buffer = NULL,
/*.mem_size =*/ ctx_size,
/*.mem_buffer =*/ NULL,
};
model.ctx = ggml_init(params);
@ -546,12 +546,13 @@ bool llama_eval(
}
struct ggml_init_params params = {
.mem_size = buf_size,
.mem_buffer = buf,
/*.mem_size =*/ buf_size,
/*.mem_buffer =*/ buf,
};
struct ggml_context * ctx0 = ggml_init(params);
struct ggml_cgraph gf = { .n_threads = n_threads };
ggml_cgraph gf = {};
gf.n_threads = n_threads;
struct ggml_tensor * embd = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
memcpy(embd->data, embd_inp.data(), N*ggml_element_size(embd));
@ -733,6 +734,7 @@ bool llama_eval(
}
int main(int argc, char ** argv) {
ggml_time_init();
const int64_t t_main_start_us = ggml_time_us();
gpt_params params;