增加对商汤科技商汤商量系列模型的支持（已经和lagent项目一起测试OK） Add support for SenseTime's SenseNova series LLM (Has been tested OK with lagent project)

README.md

@@ -53,7 +53,7 @@ pip install -r requirements.txt
 Setup FastAPI Server.
 
 ```bash
-python -m mindsearch.app --lang en --model_format internlm_server --search_engine DuckDuckGoSearch
+python -m mindsearch.app --lang en --model_format sensenova --search_engine DuckDuckGoSearch
 ```
 
 - `--lang`: language of the model, `en` for English and `cn` for Chinese.

frontend/React/vite.config.ts

@@ -53,10 +53,11 @@ export default defineConfig({
   server: {
     port: 8080,
     proxy: {
-      // "/solve": {
-      //   target: "https://mindsearch.openxlab.org.cn",
-      //   changeOrigin: true,
-      // },
+      "/solve": {
+        target: "http://127.0.0.1:8002",
+        changeOrigin: true,
+        rewrite: (path) => path.replace(/^\/solve/, '/solve'),
+      },
     },
   },
 });

frontend/mindsearch_streamlit.py

@@ -317,3 +317,6 @@ def main():
 
 if __name__ == '__main__':
     main()
+
+# 指定端口为7860的运行方法
+# streamlit run frontend/mindsearch_streamlit.py --server.port=7860

mindsearch/agent/models.py

@@ -1,6 +1,6 @@
 import os
 
-from lagent.llms import (GPTAPI, INTERNLM2_META, HFTransformerCasualLM,
+from lagent.llms import (SENSENOVA_API, GPTAPI, INTERNLM2_META, HFTransformerCasualLM,
                          LMDeployClient, LMDeployServer)
 
 internlm_server = dict(type=LMDeployServer,
@@ -36,9 +36,24 @@
                    stop_words=['<|im_end|>'])
 # openai_api_base needs to fill in the complete chat api address, such as: https://api.openai.com/v1/chat/completions
 gpt4 = dict(type=GPTAPI,
-            model_type='gpt-4-turbo',
+            model_type='gpt-4o-mini',
             key=os.environ.get('OPENAI_API_KEY', 'YOUR OPENAI API KEY'),
             openai_api_base=os.environ.get('OPENAI_API_BASE', 'https://api.openai.com/v1/chat/completions'),
+            max_new_tokens=24576,
+            )
+
+# First, apply for SenseNova's ak and sk from SenseTime staff
+# Then, generated SENSENOVA_API_KEY using lagent.utils.gen_key.auto_gen_jwt_token(ak, sk) here
+# https://github.com/InternLM/lagent/blob/ffc4ca71b4bcdbfb3a69bc0dccfa2dcc584a474d/lagent/utils/gen_key.py#L23
+
+# If you want to switch to the locally deployed SenseNova model, you need to add the model name and context-window-length here
+# https://github.com/winer632/lagent/blob/a5284a9af4c373a3ac666c51d6cef6de1e1de509/lagent/llms/sensenova.py#L21
+# You also need to change the SENSENOVA_API_BASE environment variable to the API address of the local inference framework
+sensenova = dict(type=SENSENOVA_API,
+            model_type='SenseChat-5',
+            key=os.environ.get('SENSENOVA_API_KEY', 'YOUR SENSENOVA API KEY'),
+            sensenova_api_base=os.environ.get('SENSENOVA_API_BASE', 'https://api.sensenova.cn/v1/llm/chat-completions'),
+            max_new_tokens=24576,
             )
 
 url = 'https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation'

requirements.txt

@@ -1,10 +1,10 @@
 duckduckgo_search==5.3.1b1
 einops
 fastapi
-git+https://github.com/InternLM/lagent.git
+xlagent==0.2.1
 gradio
 janus
-lmdeploy
+git+https://github.com/InternLM/lmdeploy.git
 pyvis
 sse-starlette
 termcolor