feat: add args to allow user can control the communication datatype

2025-08-23 13:43:14 +00:00
parent b780a35577
commit 3a97a82f52
5 changed files with 88 additions and 33 deletions
@@ -2118,6 +2118,13 @@ gpt_params_context gpt_params_parser_init(gpt_params & params, llama_example ex,
            params.enable_comm_compute_log = true;
        }
    ));
+    add_opt(llama_arg(
+        {"--comm-datatype"}, "TYPE",
+        format("Datatype for communication, currently support f32, q8_0, q4_0 (default: %s)", params.comm_datatype.c_str()),
+        [](gpt_params & params, const std::string & value) {
+            params.comm_datatype = value;
+        }
+    ));
    add_opt(llama_arg(
        {"--positive-file"}, "FNAME",
        format("positive prompts file, one prompt per line (default: '%s')", params.cvector_positive_file.c_str()),