Update README.md

llm/chatglm_cpp/README.md

@@ -62,6 +62,18 @@ python ./llm/cpp/convert_tokenizers.py ./build/thirdparty/openvino_contrib/modul
 ```
 
 4. Run
+
+Paramters:
+* `-m, --model` PATH        Chatglm OpenVINO model path (default: openvino_model.xml)\n"
+* `-token` PATH             Tokenizer model path (tokenizer.xml)\n"
+* `-detoken` PATH           DeTokenizer model path (detokenizer.xml)\n"
+* `-d, --device`            Device (default: GPU)\n"
+* `--convert_kv_fp16`       Convert kvcache fp16 (default: False)\n"
+* `--do_sample`             Search (default: False)\n"
+* `--top_k` N               top-k sampling (default: 0)\n"
+* `--top_p` N               top-p sampling (default: 0.7)\n"
+* `--temp` N                temperature (default: 0.95)\n"
+* `--repeat_penalty` N      penalize repeat sequence of tokens (default: 1.0, 1.0 = disabled)\n";
 
 For Linux