llama.cpp on a jailbroken PS5: run any GGUF model natively on the console, starting with Qwen 3.8
App: list models on every drive the downloads can go to (#44)
Mert Cobanova95cb0f36c4a
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex d63159a..3e9b620 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -21,6 +21,12 @@ v0.1.x are the upstream payload releases by cobanov.compute graph with the eval callback that llama.cpp already has, so no work is repeated and norequest pays for it unless it asks.+### Fixed+- Models saved to an M.2 or USB drive show up in the library wherever that drive is mounted. The+ library looked only in `/mnt/ext0`, `/mnt/usb0` and `/mnt/usb1`, while downloads can go to any+ drive the exporter reports; it now also reads the download folder and every mounted drive's+ `PS5LM/models` (#43).+## [0.9.1] - 2026-10-20### Fixeddiff --git a/ps5/app/main.cpp b/ps5/app/main.cppindex 2b9ebc4..6bf8097 100644--- a/ps5/app/main.cpp+++ b/ps5/app/main.cpp@@ -480,7 +480,19 @@ struct Models {static std::vector<ModelRow> plan_rows(const Settings & settings, double budget_gib, const std::string & running) {std::vector<ModelRow> rows;const char * kv = settings.kv_type == "auto" ? nullptr : settings.kv_type.c_str();- for (const auto & path : find_models(kModelDirs)) {+ // Also every folder the downloads can go to: the M.2 slot or a USB drive is+ // not always ext0, usb0 or usb1 (#43).+ std::vector<std::string> dirs = kModelDirs;+ std::vector<std::string> saves = { settings.download_dir };+ for (const auto & d : mounted_drives()) {+ saves.push_back(d.models);+ }+ for (const auto & d : saves) {+ if (std::find(dirs.begin(), dirs.end(), d) == dirs.end()) {+ dirs.push_back(d);+ }+ }+ for (const auto & path : find_models(dirs)) {const ModelInfo m = read_model_info(path);ModelRow row;row.file = path;