Update app.py

app.py

@@ -15,12 +15,12 @@ from io import BytesIO
 
 from serve.conversation import (default_conversation, conv_templates, SeparatorStyle)
 from serve.constants import LOGDIR
-from serve.utils import (build_logger, server_error_msg, violates_moderation, moderation_msg)
+from serve.utils import (build_#logger, server_error_msg, violates_moderation, moderation_msg)
 import subprocess
 
 subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
 
-logger = build_logger("gradio_web_server", "gradio_web_server.log")
+#logger = build_#logger("gradio_web_server", "gradio_web_server.log")
 
 headers = {"User-Agent": "Bunny Client"}
 
@@ -82,7 +82,7 @@ def get_model_list():
     ret = requests.post(args.controller_url + "/list_models")
     models = ret.json()["models"]
     models.sort(key=lambda x: priority.get(x, x))
-    logger.info(f"Models: {models}")
+    #logger.info(f"Models: {models}")
     return models
 
 
@@ -97,7 +97,7 @@ function() {
 
 
 def load_demo(url_params, request: gr.Request):
-    logger.info(f"load_demo. ip: {request.client.host}. params: {url_params}")
+    #logger.info(f"load_demo. ip: {request.client.host}. params: {url_params}")
 
     dropdown_update = gr.update(visible=True)
     if "model" in url_params:
@@ -111,7 +111,7 @@ def load_demo(url_params, request: gr.Request):
 
 
 def load_demo_refresh_model_list(request: gr.Request):
-    logger.info(f"load_demo. ip: {request.client.host}")
+    #logger.info(f"load_demo. ip: {request.client.host}")
     models = get_model_list()
     state = default_conversation.copy()
     dropdown_update = gr.update(
@@ -134,25 +134,25 @@ def vote_last_response(state, vote_type, model_selector, request: gr.Request):
 
 
 def upvote_last_response(state, model_selector, request: gr.Request):
-    logger.info(f"upvote. ip: {request.client.host}")
+    #logger.info(f"upvote. ip: {request.client.host}")
     vote_last_response(state, "upvote", model_selector, request)
     return ("",) + (disable_btn,) * 3
 
 
 def downvote_last_response(state, model_selector, request: gr.Request):
-    logger.info(f"downvote. ip: {request.client.host}")
+    #logger.info(f"downvote. ip: {request.client.host}")
     vote_last_response(state, "downvote", model_selector, request)
     return ("",) + (disable_btn,) * 3
 
 
 def flag_last_response(state, model_selector, request: gr.Request):
-    logger.info(f"flag. ip: {request.client.host}")
+    #logger.info(f"flag. ip: {request.client.host}")
     vote_last_response(state, "flag", model_selector, request)
     return ("",) + (disable_btn,) * 3
 
 
 def regenerate(state, image_process_mode, request: gr.Request):
-    logger.info(f"regenerate. ip: {request.client.host}")
+    #logger.info(f"regenerate. ip: {request.client.host}")
     state.messages[-1][-1] = None
     prev_human_msg = state.messages[-2]
     if type(prev_human_msg[1]) in (tuple, list):
@@ -162,7 +162,7 @@ def regenerate(state, image_process_mode, request: gr.Request):
 
 
 def clear_history(request: gr.Request):
-    logger.info(f"clear_history. ip: {request.client.host}")
+    #logger.info(f"clear_history. ip: {request.client.host}")
     state = default_conversation.copy()
     return (state, state.to_gradio_chatbot(), "", None) + (disable_btn,) * 5
 
@@ -196,7 +196,7 @@ def save_conversation(conversation):
 
 
 def add_text(state, text, image, image_process_mode, request: gr.Request):
-    logger.info(f"add_text. ip: {request.client.host}. len: {len(text)}")
+    #logger.info(f"add_text. ip: {request.client.host}. len: {len(text)}")
     if len(text) <= 0 and image is None:
         state.skip_next = True
         return (state, state.to_gradio_chatbot(), "", None) + (no_change_btn,) * 5
@@ -216,7 +216,7 @@ def add_text(state, text, image, image_process_mode, request: gr.Request):
         text = (text, image, image_process_mode)
         if len(state.get_images(return_pil=True)) > 0:
             state = default_conversation.copy()
-    logger.info(f"Input Text: {text}")
+    #logger.info(f"Input Text: {text}")
     state.append_message(state.roles[0], text)
     state.append_message(state.roles[1], None)
     state.skip_next = False
@@ -224,7 +224,7 @@ def add_text(state, text, image, image_process_mode, request: gr.Request):
 
 
 def http_bot(state, model_selector, temperature, top_p, max_new_tokens, request: gr.Request):
-    logger.info(f"http_bot. ip: {request.client.host}")
+    #logger.info(f"http_bot. ip: {request.client.host}")
     start_tstamp = time.time()
     model_name = model_selector
 
@@ -240,13 +240,13 @@ def http_bot(state, model_selector, temperature, top_p, max_new_tokens, request:
         new_state.append_message(new_state.roles[1], None)
         state = new_state
 
-    logger.info(f"Processed Input Text: {state.messages[-2][1]}")
+    #logger.info(f"Processed Input Text: {state.messages[-2][1]}")
     # Query worker address
     controller_url = args.controller_url
     ret = requests.post(controller_url + "/get_worker_address",
                         json={"model": model_name})
     worker_addr = ret.json()["address"]
-    logger.info(f"model_name: {model_name}, worker_addr: {worker_addr}")
+    #logger.info(f"model_name: {model_name}, worker_addr: {worker_addr}")
 
     # No available worker
     if worker_addr == "":
@@ -276,7 +276,7 @@ def http_bot(state, model_selector, temperature, top_p, max_new_tokens, request:
          "stop": '<|im_end|>', #state.sep if state.sep_style in [SeparatorStyle.PLAIN, ] else state.sep2,
         "images": f'List of {len(state.get_images())} images: {all_image_hash}',
     }
-    logger.info(f"==== request ====\n{pload}")
+    #logger.info(f"==== request ====\n{pload}")
 
     pload['images'] = state.get_images()
     print('=========> get_images')
@@ -313,7 +313,7 @@ def http_bot(state, model_selector, temperature, top_p, max_new_tokens, request:
     yield (state, state.to_gradio_chatbot()) + (enable_btn,) * 5
 
     finish_tstamp = time.time()
-    logger.info(f"{output}")
+    #logger.info(f"{output}")
 
     with open(get_conv_log_filename(), "a") as fout:
         data = {
@@ -519,14 +519,15 @@ if __name__ == "__main__":
     parser.add_argument("--concurrency-count", type=int, default=10)
     parser.add_argument("--model-list-mode", type=str, default="once",
                         choices=["once", "reload"])
+    parser.add_argument("--controller-url", type=str, default="http://localhost:10000")
     parser.add_argument("--share", action="store_true")
     parser.add_argument("--moderate", action="store_true")
     parser.add_argument("--embed", action="store_true")
     args = parser.parse_args()
-    logger.info(f"args: {args}")
+    #logger.info(f"args: {args}")
 
     models = get_model_list()
-    logger.info(args)
+    #logger.info(args)
 
     concurrency_count = int(os.getenv("concurrency_count", 5))