Rename argument

common/common.cpp

@@ -1460,12 +1460,12 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
         params.chat_template = argv[i];
         return true;
     }
-    if (arg == "--lcp-similarity") {
+    if (arg == "--slot-prompt-similarity" || arg == "-sps") {
         if (++i >= argc) {
             invalid_param = true;
             return true;
         }
-        params.lcp_similarity = std::stof(argv[i]);
+        params.slot_prompt_similarity = std::stof(argv[i]);
         return true;
     }
     if (arg == "-pps") {
@@ -1839,8 +1839,8 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
                                                                         "set custom jinja chat template (default: template taken from model's metadata)\n"
                                                                         "only commonly used templates are accepted:\n"
                                                                         "https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template" });
-    options.push_back({ "server",      "       --lcp-similarity SIMILARITY",
-                                                                        "how much the prompt of a request must match the prompt of a slot in order to use that slot (default: %.2f)\n", params.lcp_similarity });
+    options.push_back({ "server",      "-sps,  --slot-prompt-similarity SIMILARITY",
+                                                                        "how much the prompt of a request must match the prompt of a slot in order to use that slot (default: %.2f, 0.0 = disabled)\n", params.slot_prompt_similarity });
 
 #ifndef LOG_DISABLE_LOGS
     options.push_back({ "logging" });

common/common.h

@@ -202,7 +202,7 @@ struct gpt_params {
 
     std::string slot_save_path;
 
-    float lcp_similarity = 0.0f;
+    float slot_prompt_similarity = 0.5f;
 
     // batched-bench params
     bool is_pp_shared = false;

examples/server/server.cpp

@@ -647,8 +647,8 @@ struct server_context {
 
     server_metrics metrics;
 
-    // Longest Common Prefix similarity for slot selection
-    float lcp_similarity = 0.0f;
+    // Necessary similarity of prompt for slot selection
+    float slot_prompt_similarity = 0.0f;
 
     ~server_context() {
         if (ctx) {
@@ -812,7 +812,7 @@ struct server_context {
         server_slot * ret = nullptr;
 
         // find the slot that has at least n% prompt similarity
-        if (ret == nullptr && lcp_similarity != 0.0f && !prompt.empty()) {
+        if (ret == nullptr && slot_prompt_similarity != 0.0f && !prompt.empty()) {
             int max_lcp_len = 0;
             float similarity = 0;
 
@@ -840,7 +840,7 @@ struct server_context {
                 similarity = static_cast<float>(lcp_len) / slot_prompt_len;
 
                 // select the current slot if the criteria match
-                if (lcp_len > max_lcp_len && similarity > lcp_similarity) {
+                if (lcp_len > max_lcp_len && similarity > slot_prompt_similarity) {
                     max_lcp_len = lcp_len;
                     ret = &slot;
                 }
@@ -2568,8 +2568,8 @@ int main(int argc, char ** argv) {
         log_data["api_key"] = "api_key: " + std::to_string(params.api_keys.size()) + " keys loaded";
     }
 
-    // Longest Common Prefix similarity for slot selection
-    ctx_server.lcp_similarity = params.lcp_similarity;
+    // Necessary similarity of prompt for slot selection
+    ctx_server.slot_prompt_similarity = params.slot_prompt_similarity;
 
     // load the model
     if (!ctx_server.load_model(params)) {