vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/docker.yml
#	README.md
#	build-xcframework.sh
#	examples/llava/CMakeLists.txt
#	examples/llava/clip.cpp
#	examples/rpc/rpc-server.cpp
#	examples/run/run.cpp
#	ggml/src/ggml-cann/ggml-cann.cpp
#	scripts/sync-ggml-am.sh
#	scripts/sync-ggml.last
#	tests/test-backend-ops.cpp
#	tests/test-chat.cpp
This commit is contained in:
Concedo 2025-04-12 10:06:47 +08:00
commit a0ae187563
33 changed files with 2247 additions and 1202 deletions
Download patch file Download diff file Expand all files Collapse all files

55
ggml/src/ggml-cpu/ggml-cpu.c
View file

@ -2041,41 +2041,6 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
{
ggml_compute_forward_rwkv_wkv7(params, tensor);
} break;
case GGML_OP_MAP_UNARY:
{
ggml_unary_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_unary(params, tensor, fun);
}
break;
case GGML_OP_MAP_BINARY:
{
ggml_binary_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_binary(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM1_F32:
{
ggml_custom1_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_custom1_f32(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM2_F32:
{
ggml_custom2_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_custom2_f32(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM3_F32:
{
ggml_custom3_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_custom3_f32(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM1:
{
ggml_compute_forward_map_custom1(params, tensor);
@ -2091,6 +2056,11 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
ggml_compute_forward_map_custom3(params, tensor);
}
break;
case GGML_OP_CUSTOM:
{
ggml_compute_forward_custom(params, tensor);
}
break;
case GGML_OP_CROSS_ENTROPY_LOSS:
{
ggml_compute_forward_cross_entropy_loss(params, tensor);
@ -2342,11 +2312,6 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
case GGML_OP_WIN_PART:
case GGML_OP_WIN_UNPART:
case GGML_OP_GET_REL_POS:
case GGML_OP_MAP_UNARY:
case GGML_OP_MAP_BINARY:
case GGML_OP_MAP_CUSTOM1_F32:
case GGML_OP_MAP_CUSTOM2_F32:
case GGML_OP_MAP_CUSTOM3_F32:
{
n_tasks = 1;
} break;
@ -2380,6 +2345,16 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
n_tasks = MIN(p.n_tasks, n_threads);
}
} break;
case GGML_OP_CUSTOM:
{
struct ggml_custom_op_params p;
memcpy(&p, node->op_params, sizeof(p));
if (p.n_tasks == GGML_N_TASKS_MAX) {
n_tasks = n_threads;
} else {
n_tasks = MIN(p.n_tasks, n_threads);
}
} break;
case GGML_OP_CROSS_ENTROPY_LOSS:
case GGML_OP_CROSS_ENTROPY_LOSS_BACK:
case GGML_OP_OPT_STEP_ADAMW: