Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

common: refactor model handling (#24980)

Browse Source 
* common: refactor models handling

* remote preset

* cont

* rm skip_download option

* missing header

* fix plan.model_files

* fix --offline case

* move hf_plan to download

* refactor

* rm redundant curr_ex, add comments

* adapt
This commit is contained in:
Xuan-Son Nguyen
2026-06-25 15:17:51 +02:00
committed by GitHub
parent e8ecce53b8
commit 60bc8866b1
9 changed files with 289 additions and 327 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/server/server.cpp
+15 -9
View File
@@ -89,15 +89,16 @@ int llama_server(int argc, char ** argv) {
llama_backend_init();
llama_numa_init(params.numa);
// note: router mode also accepts -hf remote-preset, so we need to check that first
if (!params.model.hf_repo.empty()) {
try {
common_params_handle_models_params handle_params;
handle_params.preset_only = true;
common_params_handle_models(params, LLAMA_EXAMPLE_SERVER, handle_params);
} catch (const std::exception & e) {
// ignored for now
common_models_handler models_handler;
try {
models_handler = common_models_handler_init(params, LLAMA_EXAMPLE_SERVER);
if (common_models_handler_is_preset_repo(models_handler)) {
// apply the preset and start the server in router mode
common_models_handler_apply(models_handler, params);
}
} catch (const std::exception & e) {
SRV_ERR("failed to fetch model metadata: %s\n", e.what());
return 1;
}
// router server never loads a model and must not touch the GPU
@@ -274,7 +275,12 @@ int llama_server(int argc, char ** argv) {
return child.run_download(params);
} else if (!is_router_server) {
// single-model mode (NOT spawned by router)
common_params_handle_models(params, LLAMA_EXAMPLE_SERVER, {});
try {
common_models_handler_apply(models_handler, params);
} catch (const std::exception & e) {
SRV_ERR("failed to download model: %s\n", e.what());
return 1;
}
}
//