Update app.py

app.py

@@ -1,25 +1,42 @@
+import os
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/hf_cache"
+
 from flask import Flask, request, jsonify
-from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+import torch
 
 app = Flask(__name__)
 
-model_id = "memorease/base-t5-v2"
-tokenizer = AutoTokenizer.from_pretrained(model_id, use_fast=False)
-model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
+# 🔁 Yeni model ismi burada
+model_name = "memorease/base-t5-v2"
+
+print("[Startup] Loading model...")
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
+print("[Startup] Model loaded.")
 
 @app.route("/ask", methods=["POST"])
 def ask_question():
-    data = request.json
-    memory = data["text"]
+    try:
+        input_text = request.json.get("text")
+        if not input_text:
+            return jsonify({"error": "Missing 'text'"}), 400
+
+        prompt = f"Only generate a factual and relevant question about this memory: {input_text}"
+        inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True)
+
+        with torch.no_grad():
+            outputs = model.generate(**inputs, max_new_tokens=64)
 
-    instruction = "Generate a question based on the given memory."
-    prompt = instruction + " " + memory
+        question = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        return jsonify({"question": question})
 
-    inputs = tokenizer(prompt, return_tensors="pt", truncation=True, padding=True)
-    outputs = model.generate(**inputs, max_new_tokens=64)
-    result = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
 
-    return jsonify(question=result)
+@app.route("/", methods=["GET"])
+def healthcheck():
+    return jsonify({"status": "running"})
 
 if __name__ == "__main__":
     app.run(host="0.0.0.0", port=7860)