Warm up backend by making dummy requests for both OpenAI and SpaCy

backend/nlp.py

@@ -7,6 +7,7 @@
 from pydantic import BaseModel
 import spacy
 
+import openai
 from openai.types.chat.chat_completion_message_param import ChatCompletionMessageParam
 from openai import AsyncOpenAI
 
@@ -37,6 +38,23 @@
     exit()
 
 
+async def warmup_nlp():
+    # Warm up the OpenAI client by making a dummy request
+    dummy_client = openai.AsyncOpenAI(base_url="https://localhost:8000/v1", api_key="", timeout=0.01, max_retries=0)
+    # make a dummy request to make sure everything is imported
+    try:
+        await dummy_client.chat.completions.create(
+            model=MODEL_NAME,
+            messages=[{"role": "user", "content": "Hello"}],
+        )
+    except openai.APIConnectionError:
+        # We expect this error because we're connecting to a non-existent server
+        pass
+
+
+    # Warm up the SpaCy model by processing a sample text
+    nlp("Hello world. This is a test.").sents
+
 prompts = {
     "example_sentences": """\
 We're helping a writer draft a document. Please output three possible options for inspiring and fresh possible next sentences that would help the writer think about what they should write next. Guidelines:

backend/server.py

@@ -3,6 +3,7 @@
 import logging
 import os
 import zipfile
+from contextlib import asynccontextmanager
 from datetime import datetime
 from pathlib import Path
 from typing import Annotated, Dict, List, Literal, Optional
@@ -112,8 +113,12 @@ class ReflectionLog(Log):
     paragraph: str
 
 
-# Initialize Server
-app = FastAPI()
+@asynccontextmanager
+async def app_lifespan(app: FastAPI):
+    await nlp.warmup_nlp()
+    yield
+
+app = FastAPI(lifespan=app_lifespan)
 
 origins = ["*"]
 

frontend/src/api/index.ts

@@ -7,33 +7,6 @@ export interface LogPayload {
 	[key: string]: any;
 }
 
-export async function pingServer(): Promise<void> {
-  try {
-		// Ping the server, which returns the server's current timestamp
-		const url = `${SERVER_URL}/ping`;
-    const response = await fetch(url, {
-      method: 'GET',
-      headers: {
-        'Content-Type': 'application/json',
-      }
-    });
-
-    if (!response.ok) {
-			// eslint-disable-next-line no-console
-      console.warn('Server ping failed:', response.status);
-    }
-
-		const data = await response.json();
-		// eslint-disable-next-line no-console
-		console.log('Server ping successful:', data);
-  }
-	catch (error) {
-		// eslint-disable-next-line no-console
-    console.warn('Server ping error:', error);
-    // Don't throw - we want to silently handle ping failures
-  }
-}
-
 export function log(payload: LogPayload) {
 	const payloadWithTimestamp = {
 		...payload,