ExternalVendorCode/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-06-14 12:58:10 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

ggml : sync (ggml-alloc, GPU, eps, etc.) (#1220)

Browse Source 
* ggml : sync (ggml-alloc, GPU, eps, etc.)

* ggml : fix build

* wasm : fix build
This commit is contained in:
Georgi Gerganov
2023-09-05 13:54:40 +03:00
committed by GitHub
parent 6780c98e19
commit 59a3d0cb57
17 changed files with 11254 additions and 4095 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
examples/talk/gpt-2.cpp
View File

@ -379,6 +379,7 @@ bool gpt2_model_load(const std::string & fname, gpt2_model & model, gpt_vocab &
// - embd_inp: the embeddings of the tokens in the context
// - embd_w: the predicted logits for the next token
//
// TODO: sync latest version from ggml repo
bool gpt2_eval(
const gpt2_model & model,
const int n_threads,
@ -420,7 +421,6 @@ bool gpt2_eval(
struct ggml_context * ctx0 = ggml_init(params);
struct 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));
@ -442,7 +442,7 @@ bool gpt2_eval(
// norm
{
// [ 768, N]
cur = ggml_norm(ctx0, inpL);
cur = ggml_norm(ctx0, inpL, 1e-5f);
// cur = ln_1_g*cur + ln_1_b
// [ 768, N]
@ -589,7 +589,7 @@ bool gpt2_eval(
{
// norm
{
cur = ggml_norm(ctx0, inpFF);
cur = ggml_norm(ctx0, inpFF, 1e-5f);
// cur = ln_2_g*cur + ln_2_b
// [ 768, N]
@ -644,7 +644,7 @@ bool gpt2_eval(
// norm
{
// [ 768, N]
inpL = ggml_norm(ctx0, inpL);
inpL = ggml_norm(ctx0, inpL, 1e-5f);
// inpL = ln_f_g*inpL + ln_f_b
// [ 768, N]
@ -664,8 +664,8 @@ bool gpt2_eval(
//inpL = ggml_soft_max(ctx0, inpL);
// run the computation
ggml_build_forward_expand(&gf, inpL);
ggml_graph_compute (ctx0, &gf);
ggml_build_forward_expand (&gf, inpL);
ggml_graph_compute_with_ctx(ctx0, &gf, n_threads);
//if (n_past%100 == 0) {
// ggml_graph_print (&gf);