Merge branch 'main' into gradio-multi

fastchat/serve/gradio_web_server.py

@@ -136,14 +136,8 @@ def get_conv_log_filename():
     return name
 
 
-def get_model_list(
-    controller_url,
-    register_openai_compatible_models,
-    add_chatgpt,
-    add_claude,
-    add_palm,
-    multimodal,
-):
+def get_model_list(controller_url, register_api_endpoint_file, multimodal):
+    global api_endpoint_info
     if controller_url:
         ret = requests.post(controller_url + "/refresh_all_workers")
         assert ret.status_code == 200
@@ -204,11 +198,8 @@ def load_demo(url_params, request: gr.Request):
     if args.model_list_mode == "reload":
         models, all_models = get_model_list(
             controller_url,
-            args.register_openai_compatible_models,
-            args.add_chatgpt,
-            args.add_claude,
-            args.add_palm,
-            False,
+            args.register_api_endpoint_file,
+            False
         )
 
     return load_demo_single(models, url_params)
@@ -966,6 +957,7 @@ def build_demo(models):
     models, all_models = get_model_list(
         args.controller_url,
         args.register_api_endpoint_file,
+        False
     )
 
     # Set authorization credentials

fastchat/serve/gradio_web_server_multi.py

@@ -47,7 +47,7 @@
 
 
 def load_demo(url_params, request: gr.Request):
-    global models, vision_language_models
+    global models, all_models, vision_language_models
 
     ip = get_ip(request)
     logger.info(f"load_demo. ip: {ip}. params: {url_params}")
@@ -66,31 +66,15 @@ def load_demo(url_params, request: gr.Request):
         selected = 4
 
     if args.model_list_mode == "reload":
-        if args.anony_only_for_proprietary_model:
-            models = get_model_list(
-                args.controller_url,
-                args.register_openai_compatible_models,
-                False,
-                False,
-                False,
-                False,
-            )
-        else:
-            models = get_model_list(
-                args.controller_url,
-                args.register_openai_compatible_models,
-                args.add_chatgpt,
-                args.add_claude,
-                args.add_palm,
-                False,
-            )
-
-        vision_language_models = get_model_list(
+        models, all_models = get_model_list(
             args.controller_url,
-            args.register_openai_compatible_models,
-            False,
-            False,
+            args.register_api_endpoint_file,
             False,
+        )
+
+        vision_language_models, _ = get_model_list(
+            args.controller_url,
+            args.register_api_endpoint_file,
             True,
         )
 
@@ -232,31 +216,15 @@ def build_demo(
     set_global_vars(args.controller_url, args.moderate)
     set_global_vars_named(args.moderate)
     set_global_vars_anony(args.moderate)
-    if args.anony_only_for_proprietary_model:
-        models = get_model_list(
-            args.controller_url,
-            args.register_openai_compatible_models,
-            False,
-            False,
-            False,
-            False,
-        )
-    else:
-        models = get_model_list(
-            args.controller_url,
-            args.register_openai_compatible_models,
-            args.add_chatgpt,
-            args.add_claude,
-            args.add_palm,
-            False,
-        )
-
-    vision_language_models = get_model_list(
+    models, all_models = get_model_list(
         args.controller_url,
-        args.register_openai_compatible_models,
-        False,
-        False,
+        args.register_api_endpoint_file,
         False,
+    )
+
+    vision_language_models, _ = get_model_list(
+        args.controller_url,
+        args.register_api_endpoint_file,
         True,
     )