llama : auto download HF models if URL provided

Author: ggerganov

llama.cpp

@@ -3909,9 +3909,57 @@ static bool llm_load_tensors(
     return true;
 }
 
+// check if the URL is a HuggingFace model, and if so, try to download it
+static void hf_try_download_model(std::string & url) {
+    bool is_url = false;
+
+    if (url.size() > 22) {
+        is_url = (url.compare(0, 22, "https://huggingface.co") == 0);
+    }
+
+    if (!is_url) {
+        return;
+    }
+
+    // Examples:
+    //
+    // https://huggingface.co/TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF/resolve/main/mixtral-8x7b-instruct-v0.1.Q2_K.gguf
+
+    std::string basename;
+    basename = url.substr(url.find_last_of("/\\") + 1);
+
+    LLAMA_LOG_INFO("%s: detected URL, attempting to download %s\n", __func__, basename.c_str());
+
+    {
+        const std::string cmd = "wget -q --show-progress -c -O " + basename + " " + url;
+        LLAMA_LOG_INFO("%s: %s\n", __func__, cmd.c_str());
+
+        const int ret = system(cmd.c_str());
+        if (ret == 0) {
+            url = basename;
+            return;
+        }
+    }
+
+    {
+        const std::string cmd = "curl -C - -f -o " + basename + " -L " + url;
+        LLAMA_LOG_INFO("%s: %s\n", __func__, cmd.c_str());
+
+        const int ret = system(cmd.c_str());
+        if (ret == 0) {
+            url = basename;
+            return;
+        }
+    }
+
+    LLAMA_LOG_WARN("%s: failed to download\n", __func__);
+}
+
 // Returns 0 on success, -1 on error, and -2 on cancellation via llama_progress_callback
-static int llama_model_load(const std::string & fname, llama_model & model, const llama_model_params & params) {
+static int llama_model_load(std::string fname, llama_model & model, const llama_model_params & params) {
     try {
+        hf_try_download_model(fname);
+
         llama_model_loader ml(fname, params.use_mmap, params.kv_overrides);
 
         model.hparams.vocab_only = params.vocab_only;