update

README.md

@@ -154,7 +154,7 @@ Launch inference server by:
 # install lmdeploy with extra dependencies
 pip install lmdeploy[serve]
 
-lmdeploy serve api_server internlm/internlm-chat-7b-v1_1 --model-name internlm-chat-7b --server_name 0.0.0.0 --server_port ${server_port} --instance_num 32 --tp 1
+lmdeploy serve api_server internlm/internlm-chat-7b-v1_1 --model-name internlm-chat-7b --instance_num 32 --tp 1
 ```
 
 Then, you can communicate with it by command line,

README_zh-CN.md

@@ -153,7 +153,7 @@ lmdeploy serve gradio internlm/internlm-chat-7b-v1_1 --model-name internlm-chat-
 # 安装lmdeploy额外依赖
 pip install lmdeploy[serve]
 
-lmdeploy serve api_server internlm/internlm-chat-7b-v1_1 --model-name internlm-chat-7b --server_name 0.0.0.0 --server_port ${server_port} --instance_num 32 --tp 1
+lmdeploy serve api_server internlm/internlm-chat-7b-v1_1 --model-name internlm-chat-7b --instance_num 32 --tp 1
 ```
 
 你可以通过命令行方式与推理服务进行对话：

src/turbomind/triton_backend/llama/LlamaTritonModel.cc

@@ -119,9 +119,12 @@ LlamaTritonModel<T>::LlamaTritonModel(size_t      tensor_para_size,
     enable_custom_all_reduce_(enable_custom_all_reduce)
 {
     INIReader reader;
-    bool      init_with_config = false;
+    if (!config.empty() && !model_dir.empty()) {
+        TM_LOG_ERROR("[ERROR] config and model_dir are all set");
+        ft::FT_CHECK(false);
+    }
 
-    if (!init_with_config && !config.empty()) {
+    if (!config.empty()) {
         std::FILE* tmpf = std::tmpfile();
         std::fputs(config.c_str(), tmpf);
         std::rewind(tmpf);
@@ -130,10 +133,9 @@ LlamaTritonModel<T>::LlamaTritonModel(size_t      tensor_para_size,
             TM_LOG_ERROR("[ERROR] Can't init with config %s", config.c_str());
             ft::FT_CHECK(false);
         }
-        init_with_config = true;
     }
 
-    if (!init_with_config && !model_dir.empty()) {
+    if (!model_dir.empty()) {
         model_dir_ = model_dir;
         const std::string inifile{model_dir + "/config.ini"};
         reader = INIReader(inifile);