Update app.py

app.py

@@ -1,30 +1,30 @@
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
 import gradio as gr
 
-# LLaMA 2 Chat modeli
-model_id = "meta-llama/Llama-2-7b-chat-hf"
+# Model yükleme (hafif!)
+model_id = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
 
 tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(
-    model_id,
-    torch_dtype=torch.float16,  # CPU çalışıyorsan float32 olabilir
-    device_map="auto"
-)
+model = AutoModelForCausalLM.from_pretrained(model_id)
 
 def generate_question(memory):
-    prompt = f"[INST] You are a helpful assistant. Based on this memory, generate a question that would help the user recall more details:\n\nMemory: {memory}\n\nQuestion: [/INST]"
-    inputs = tokenizer(prompt, return_tensors="pt")
-    outputs = model.generate(**inputs, max_new_tokens=50)
-    result = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    return result.split("Question:")[-1].strip()
+    prompt = f"<|system|>You are a helpful assistant that generates meaningful questions from memories.<|user|>Memory: {memory}\nGenerate a related question.<|assistant|>"
+
+    input_ids = tokenizer(prompt, return_tensors="pt").input_ids
+    output_ids = model.generate(input_ids, max_new_tokens=50, do_sample=True)
+    result = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+
+    # Cevabın son kısmını al
+    question = result.split("<|assistant|>")[-1].strip()
+    return question
 
-# Arayüz
+# Gradio UI
 iface = gr.Interface(
     fn=generate_question,
     inputs=gr.Textbox(label="Your Memory"),
     outputs=gr.Textbox(label="Generated Question"),
-    title="LLaMA Chat Question Generator"
+    title="TinyLLaMA Memory Question Generator"
 )
 
 iface.launch()