vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-07 00:59:03 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix: n_worker in draft model

Browse source 
(cherry picked from commit 921ad2b453b24b715ad5db6a703fb3df65fdcb80)
This commit is contained in:
DeEMO 2025-06-17 13:20:06 +08:00
parent 2b902f89bd
commit deeec668b8
1 changed files with 2 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
common/common.cpp
View file

@ -1682,6 +1682,7 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
cparams.n_layer_window[0] = n_layers;
mparams.n_layer_window[0] = n_layers;
llama_context_n_layer_window(lctx)[0] = n_layers;
llama_update_context_with_rankworld(lctx, 0, 1, 0, 1);
#if defined(GGML_USE_METAL) || defined(GGML_USE_CUDA)
params.n_gpu_layers = std::min((int32_t)n_layers, params.n_gpu_layers);
@ -1723,7 +1724,7 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
}
// sychronize device profile to the master node
NodeType node_type;
NodeType node_type = NodeType::NODE_TYPE_WORKER;
char is_forwarder[32] = {0};
if (my_rank == 0) {
if (auto_schedule) {