common : rename speculative.draftless_type -> speculative.type

Author: ggerganov

common/arg.cpp

@@ -3400,18 +3400,18 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     add_opt(common_arg(
         {"--spec-draftless"}, "[none|ngram-cache|ngram-simple|ngram-map-k|ngram-map-k4v]",
         string_format("type of speculative decoding to use when no draft model is provided (default: %s)\n",
-            common_speculative_type_to_str(params.speculative.draftless_type).c_str()),
+            common_speculative_type_to_str(params.speculative.type).c_str()),
         [](common_params & params, const std::string & value) {
             if (value == "none") {
-                params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NONE;
+                params.speculative.type = COMMON_SPECULATIVE_TYPE_NONE;
             } else if (value == "ngram-cache") {
-                params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NGRAM_CACHE;
+                params.speculative.type = COMMON_SPECULATIVE_TYPE_NGRAM_CACHE;
             } else if (value == "ngram-simple") {
-                params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NGRAM_SIMPLE;
+                params.speculative.type = COMMON_SPECULATIVE_TYPE_NGRAM_SIMPLE;
             } else if (value == "ngram-map-k") {
-                params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K;
+                params.speculative.type = COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K;
             } else if (value == "ngram-map-k4v") {
-                params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K4V;
+                params.speculative.type = COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K4V;
             } else {
                 throw std::invalid_argument("unknown speculative decoding type without draft model");
             }

common/common.h

@@ -273,9 +273,7 @@ struct common_params_speculative {
 
     struct common_params_model model;
 
-    // draftless:
-
-    common_speculative_type draftless_type = COMMON_SPECULATIVE_TYPE_NONE; // type of speculative decoding without a draft model
+    common_speculative_type type = COMMON_SPECULATIVE_TYPE_NONE; // type of speculative decoding
 
     uint16_t ngram_size_n     = 12; // ngram size for lookup
     uint16_t ngram_size_m     = 48; // mgram size for speculative tokens

common/speculative.cpp

@@ -355,12 +355,14 @@ struct common_speculative * common_speculative_init(
     // Compute the implementations to use based on the config and their order of preference
     std::vector<common_speculative_config> configs = {}; // list of speculative configs to try
     {
-        bool has_draft =!params.model.path.empty();
+        bool has_draft = !params.model.path.empty();
         bool has_draft_eagle3 = false; // TODO PR-18039: if params.speculative.eagle3
-        bool has_ngram_cache = (params.draftless_type == COMMON_SPECULATIVE_TYPE_NGRAM_CACHE);
-        bool has_ngram_simple = (params.draftless_type == COMMON_SPECULATIVE_TYPE_NGRAM_SIMPLE);
-        bool has_ngram_map_k = (params.draftless_type == COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K);
-        bool has_ngram_map_k4v = (params.draftless_type == COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K4V);
+
+        bool has_ngram_cache   = (params.type == COMMON_SPECULATIVE_TYPE_NGRAM_CACHE);
+        bool has_ngram_simple  = (params.type == COMMON_SPECULATIVE_TYPE_NGRAM_SIMPLE);
+        bool has_ngram_map_k   = (params.type == COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K);
+        bool has_ngram_map_k4v = (params.type == COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K4V);
+
         // In a more complex implementation we could use the same implementation but with different parameters.
         // This was initially used in PR-18471 but removed to simplify the code.
         if (has_ngram_simple) {

tools/server/server-task.cpp

@@ -77,7 +77,7 @@ json task_params::to_json(bool only_metrics) const {
             {"speculative.n_max",         speculative.n_max},
             {"speculative.n_min",         speculative.n_min},
             {"speculative.p_min",         speculative.p_min},
-            {"speculative.draftless_t",   common_speculative_type_to_str(speculative.draftless_type)},
+            {"speculative.type",          common_speculative_type_to_str(speculative.type)},
             {"speculative.ngram_size_n",  speculative.ngram_size_n},
             {"speculative.ngram_size_m",  speculative.ngram_size_m},
             {"speculative.ngram_c_rate",  speculative.ngram_check_rate},
@@ -141,7 +141,7 @@ json task_params::to_json(bool only_metrics) const {
         {"speculative.n_max",         speculative.n_max},
         {"speculative.n_min",         speculative.n_min},
         {"speculative.p_min",         speculative.p_min},
-        {"speculative.draftless_t",   common_speculative_type_to_str(speculative.draftless_type)},
+        {"speculative.type",          common_speculative_type_to_str(speculative.type)},
         {"speculative.ngram_size_n",  speculative.ngram_size_n},
         {"speculative.ngram_size_m",  speculative.ngram_size_m},
         {"speculative.ngram_c_rate",  speculative.ngram_check_rate},
@@ -253,7 +253,8 @@ task_params server_task::params_from_json_cmpl(
     params.speculative.n_min = std::max(params.speculative.n_min, 0);
     params.speculative.n_max = std::max(params.speculative.n_max, 0);
 
-    params.speculative.draftless_type   = common_speculative_type_from_name(json_value(data, "speculative.draftless_t", common_speculative_type_to_str(defaults.speculative.draftless_type)));
+    params.speculative.type = common_speculative_type_from_name(json_value(data, "speculative.type", common_speculative_type_to_str(defaults.speculative.type)));
+
     params.speculative.ngram_size_n     = json_value(data, "speculative.ngram_size_n", defaults.speculative.ngram_size_n);
     params.speculative.ngram_size_m     = json_value(data, "speculative.ngram_size_m", defaults.speculative.ngram_size_m);
     params.speculative.ngram_check_rate = json_value(data, "speculative.ngram_c_rate", defaults.speculative.ngram_check_rate);