changed to uvicorn setup for HF v16

main/api.py

@@ -29,11 +29,14 @@ class InferenceApi(LitAPI):
 
     async def _get_client(self):
         """Get or create HTTP client as needed"""
-        base_url = self.llm_config.get('base_url', 'http://localhost:8002')
+        host = self.llm_config.get('host', 'localhost')
+        port = self.llm_config.get('port', 8002)
 
-        # If it's a HF space URL, remove any port specification
-        if 'hf.space' in base_url:
-            base_url = base_url.split(':')[0]  # Take everything before any port number
+        # Construct base URL, omitting port for HF spaces
+        if 'hf.space' in host:
+            base_url = f"https://{host}"
+        else:
+            base_url = f"http://{host}:{port}"
 
         return httpx.AsyncClient(
             base_url=base_url,

main/config.yaml

@@ -5,7 +5,8 @@ server:
   max_batch_size: 1
 
 llm_server:
-  base_url: "https://teamgenki-llmserver.hf.space:7860" # The base URL of the LLM server
+  host: "teamgenki-llmserver.hf.space"
+  port: 7860  # Will be ignored for hf.space URLs
   timeout: 60.0
   api_prefix: "/api/v1"  # This will be used for route prefixing
   endpoints: