readme_20240522

lmdeploy/model.py

@@ -452,6 +452,7 @@ If a question does not make any sense, or is not factually coherent, explain why
             else:
                 ret += f'{user} {self.e_inst} '
         return ret
+
 @MODELS.register_module(name='qwen2-110b')
 @MODELS.register_module(name='qwen2-72b')
 @MODELS.register_module(name='qwen2-14b')

sugon_readme.md

@@ -196,4 +196,11 @@ class Qwen7BChat(BaseModel):
                  session_len=8192,
                  top_p=0.5,
                  top_k=40,
+```
+### 命令部分(qwen1.5)
+```shell
+convert2:
+lmdeploy convert --model_name qwen2-7b --model_path /test_data/Qwen_Qwen1.5-32B-Chat --tokenizer-path /test_data/Qwen_Qwen1.5-32B-Chat/vocab.json --model_format hf
+chat:
+lmdeploy chat turbomind --model_path workspace --tp 4
 ```
\ No newline at end of file