fix predict import error and change README

README.md

@@ -102,10 +102,13 @@ To activate it, create a `.env.local` configuration file:
 
 ```bash
 LLM_ENGINE="OPENAI"
+
 # default openai api base url is: https://api.openai.com/v1
-OPENAI_API_BASE_URL="Your OpenAI API Base URL"
-OPENAI_API_KEY="Your OpenAI API Key"
-OPENAI_API_MODEL="gpt-3.5-turbo"
+LLM_OPENAI_API_BASE_URL="Your OpenAI API Base URL"
+
+LLM_OPENAI_API_MODEL="gpt-3.5-turbo"
+
+AUTH_OPENAI_API_KEY="Your OpenAI API Key"
 ```
 
 ### Option 4: Fork and modify the code to use a different LLM system
@@ -140,11 +143,11 @@ To use Replicate, create a `.env.local` configuration file:
 ```bash
 RENDERING_ENGINE="REPLICATE"
 
-REPLICATE_API_TOKEN="Your Replicate token"
+RENDERING_REPLICATE_API_MODEL="stabilityai/sdxl"
 
-REPLICATE_API_MODEL="stabilityai/sdxl"
+RENDERING_REPLICATE_API_MODEL_VERSION="da77bc59ee60423279fd632efb4795ab731d9e3ca9705ef3341091fb989b7eaf"
 
-REPLICATE_API_MODEL_VERSION="da77bc59ee60423279fd632efb4795ab731d9e3ca9705ef3341091fb989b7eaf"
+AUTH_REPLICATE_API_TOKEN="Your Replicate token"
 ```
 
 ### Option 3: Use another SDXL API

src/app/queries/predict.ts

@@ -1,9 +1,15 @@
 "use server"
 
 import { LLMEngine } from "@/types"
-import { predictWithHuggingFace } from "./predictWithHuggingFace"
-import { predictWithOpenAI } from "./predictWithOpenAI"
 
 const llmEngine = `${process.env.LLM_ENGINE || ""}` as LLMEngine
 
-export const predict = llmEngine === "OPENAI" ? predictWithOpenAI : predictWithHuggingFace
+export const predict = async () => {
+    if (llmEngine === "OPENAI")  {
+        const module = await import("./predictWithOpenAI")  
+        return module.predictWithOpenAI
+    } else {
+        const module = await import("./predictWithHuggingFace")  
+        return module.predictWithHuggingFace
+    }
+}