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

sync : ggml (backend v2) (#3912)

Browse source 
* sync : ggml (backend v2) (wip)

* sync : migrate examples and llama.cpp to dynamic graphs (wip)

* sync : update tests + fix max op params to 64

ggml-ci

* sync : ggml-cuda

ggml-ci

* llama : fix save/load state context size

ggml-ci

* sync : try to fix build on tvOS

* sync : pass custom graph sizes in training examples

* sync : update graph copies to new ggml API

* sync : update sync-ggml.sh with new files

* scripts : fix header in sync script

* train : fix context size calculations

* llama : increase inference graph size up to 4096 nodes

* train : allocate grads for backward graphs

* train : allocate grads for gb_tmp
This commit is contained in:
Georgi Gerganov 2023-11-13 14:16:23 +02:00 • committed by GitHub
parent bb50a792ec
commit 4760e7cc0b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
22 changed files with 1994 additions and 864 deletions
Download patch file Download diff file Expand all files Collapse all files

7
tests/test-grad0.cpp
View file

@ -231,9 +231,10 @@ static bool check_gradient(
printf("GGML_N_THREADS = %d\n", n_threads);
}
struct ggml_cgraph * gf = ggml_build_forward_ctx(ctx0, f);
struct ggml_cgraph * gb = ggml_new_graph(ctx0);
*gb = *gf;
struct ggml_cgraph * gf = ggml_new_graph_custom(ctx0, GGML_DEFAULT_GRAPH_SIZE, true);
struct ggml_cgraph * gb = ggml_new_graph_custom(ctx0, GGML_DEFAULT_GRAPH_SIZE, true);
ggml_build_forward_expand(gf, f);
ggml_graph_cpy(gf, gb);
ggml_build_backward_expand(ctx0, gf, gb, false);
ggml_graph_compute_with_ctx(ctx0, gf, n_threads);