support multi process to launch llmserver, add openai proxy api.

pilot/configs/config.py

@@ -17,8 +17,9 @@ class Config(metaclass=Singleton):
     def __init__(self) -> None:
         """Initialize the Config class"""
 
-        # Gradio language version: en, cn
+        # Gradio language version: en, zh
         self.LANGUAGE = os.getenv("LANGUAGE", "en")
+        self.WEB_SERVER_PORT = int(os.getenv("WEB_SERVER_PORT", 7860))
 
         self.debug_mode = False
         self.skip_reprompt = False

pilot/server/llmserver.py

@@ -84,6 +84,11 @@ class ModelWorker:
         return get_embeddings(self.model, self.tokenizer, prompt)
 
 
+model_path = LLM_MODEL_CONFIG[CFG.LLM_MODEL]
+worker = ModelWorker(
+    model_path=model_path, model_name=CFG.LLM_MODEL, device=DEVICE, num_gpus=1
+)
+
 app = FastAPI()
 
 
@@ -157,11 +162,4 @@ def embeddings(prompt_request: EmbeddingRequest):
 
 
 if __name__ == "__main__":
-    model_path = LLM_MODEL_CONFIG[CFG.LLM_MODEL]
-    print(model_path, DEVICE)
-
-    worker = ModelWorker(
-        model_path=model_path, model_name=CFG.LLM_MODEL, device=DEVICE, num_gpus=1
-    )
-
     uvicorn.run(app, host="0.0.0.0", port=CFG.MODEL_PORT, log_level="info")

pilot/server/webserver.py

@@ -658,7 +658,7 @@ def signal_handler(sig, frame):
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
     parser.add_argument("--host", type=str, default="0.0.0.0")
-    parser.add_argument("--port", type=int)
+    parser.add_argument("--port", type=int, default=CFG.WEB_SERVER_PORT)
     parser.add_argument("--concurrency-count", type=int, default=10)
     parser.add_argument(
         "--model-list-mode", type=str, default="once", choices=["once", "reload"]