Update server.js

server.js

@@ -51,27 +51,31 @@ async function initModel() {
 ----------------------- */
 
 app.post("/generate", async (req, res) => {
-    // 1. Check if the model is still loading in the background
     if (!isModelReady) {
         return res.status(503).json({ 
             error: "The AI model is still loading into memory. Please wait a few seconds and try again." 
         });
     }
 
+    let sequence; // Define this here so we can clean it up in the 'finally' block
+
     try {
         const {
             user_input,
-            user_temp = 0.7,
-            user_inst = "You are an AI assistant. Give short clear answers.",
-            user_max_token = 512
+            user_temp = 0.2,
+            user_inst = "You are an Wrld-Gpt AI assistant. Give short clear answers. Do not make assumptions",
+            user_max_token = 1024
         } = req.body;
 
         if (!user_input) {
             return res.status(400).json({ error: "Missing required field: user_input" });
         }
 
+        // Grab a sequence memory slot for this specific request
+        sequence = contextInstance.getSequence();
+
         const session = new LlamaChatSession({
-            contextSequence: contextInstance.getSequence(),
+            contextSequence: sequence,
             systemPrompt: user_inst 
         });
 
@@ -88,6 +92,11 @@ app.post("/generate", async (req, res) => {
     } catch (err) {
         console.error("Error during generation:", err);
         res.status(500).json({ error: "An internal error occurred during text generation." });
+    } finally {
+        // CRITICAL FIX: Always free up the sequence slot when done, even if an error occurs!
+        if (sequence) {
+            sequence.dispose();
+        }
     }
 });
 
@@ -95,12 +104,11 @@ app.post("/generate", async (req, res) => {
    STARTUP SEQUENCE
 ----------------------- */
 
-// 1. START THE SERVER IMMEDIATELY (This satisfies Hugging Face's health check)
 app.listen(PORT, "0.0.0.0", () => {
     console.log(`✅ Web server is listening on port ${PORT}`);
     console.log(`⏳ Starting background model load...`);
     
-    // 2. LOAD THE MODEL IN THE BACKGROUND
+    //LOAD THE MODEL IN THE BACKGROUND
     initModel().catch(err => {
         console.error("Critical Failure: Failed to load the AI model.", err);
     });