kanshan/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-06-26 14:20:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

download: add option to skip_download (#23059)

Browse Source 
* download: add option to skip_download

* fix

* fix 2

* if file doesn't exist, respect skip_download flag
This commit is contained in:
Xuan-Son Nguyen
2026-05-29 16:30:55 +02:00
committed by GitHub
parent da3f990a47
commit 06d26dfdff
8 changed files with 126 additions and 83 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/server/server-models.cpp
+37 -33
View File
@@ -180,7 +180,8 @@ void server_model_meta::update_caps() {
"LLAMA_ARG_HF_REPO",
"LLAMA_ARG_HF_REPO_FILE",
});
params.offline = true; // avoid any unwanted network call during capability detection
params.offline = true;
// params.skip_download = true; // TODO: ideally, we should validate the model here, but it takes too much time
common_params_handle_models(params, LLAMA_EXAMPLE_SERVER);
if (params.mmproj.path.empty()) {
multimodal = { false, false };
@@ -371,18 +372,19 @@ void server_models::load_models() {
// FIRST LOAD: add all models, then unlock for autoloading
for (const auto & [name, preset] : final_presets) {
server_model_meta meta{
/* preset */ preset,
/* name */ name,
/* aliases */ {},
/* tags */ {},
/* port */ 0,
/* status */ SERVER_MODEL_STATUS_UNLOADED,
/* last_used */ 0,
/* args */ std::vector<std::string>(),
/* loaded_info */ {},
/* exit_code */ 0,
/* stop_timeout */ DEFAULT_STOP_TIMEOUT,
/* multimodal */ mtmd_caps{false, false},
/* preset */ preset,
/* name */ name,
/* aliases */ {},
/* tags */ {},
/* port */ 0,
/* status */ SERVER_MODEL_STATUS_UNLOADED,
/* last_used */ 0,
/* args */ std::vector<std::string>(),
/* loaded_info */ {},
/* exit_code */ 0,
/* stop_timeout */ DEFAULT_STOP_TIMEOUT,
/* multimodal */ mtmd_caps{false, false},
/* need_download */ false,
};
add_model(std::move(meta));
}
@@ -524,18 +526,19 @@ void server_models::load_models() {
for (const auto & [name, preset] : final_presets) {
if (mapping.find(name) == mapping.end()) {
server_model_meta meta{
/* preset */ preset,
/* name */ name,
/* aliases */ {},
/* tags */ {},
/* port */ 0,
/* status */ SERVER_MODEL_STATUS_UNLOADED,
/* last_used */ 0,
/* args */ std::vector<std::string>(),
/* loaded_info */ {},
/* exit_code */ 0,
/* stop_timeout */ DEFAULT_STOP_TIMEOUT,
/* multimodal */ mtmd_caps{false, false},
/* preset */ preset,
/* name */ name,
/* aliases */ {},
/* tags */ {},
/* port */ 0,
/* status */ SERVER_MODEL_STATUS_UNLOADED,
/* last_used */ 0,
/* args */ std::vector<std::string>(),
/* loaded_info */ {},
/* exit_code */ 0,
/* stop_timeout */ DEFAULT_STOP_TIMEOUT,
/* multimodal */ mtmd_caps{false, false},
/* need_download */ false,
};
add_model(std::move(meta));
newly_added.push_back(name);
@@ -1263,14 +1266,15 @@ void server_models_routes::init_routes() {
};
json model_info = json {
{"id", meta.name},
{"aliases", meta.aliases},
{"tags", meta.tags},
{"object", "model"}, // for OAI-compat
{"owned_by", "llamacpp"}, // for OAI-compat
{"created", t}, // for OAI-compat
{"status", status},
{"architecture", architecture},
{"id", meta.name},
{"aliases", meta.aliases},
{"tags", meta.tags},
{"object", "model"}, // for OAI-compat
{"owned_by", "llamacpp"}, // for OAI-compat
{"created", t}, // for OAI-compat
{"status", status},
{"architecture", architecture},
{"need_download", meta.need_download},
// TODO: add other fields, may require reading GGUF metadata
};