Create app.py

app.py

@@ -0,0 +1,68 @@
+import gradio as ui
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
+from threading import Thread
+
+# Gunakan varian Instruct agar nyambung diajak chat (bisa switch thinking mode!)
+model_id = "Qwen/Qwen3-0.6B-Instruct"
+
+print("Memuat Tokenizer & Model Qwen3-0.6B ke RAM...")
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.float32, # float32 aman & stabil untuk CPU gratisan
+    device_map="cpu"
+)
+print("Qwen3 Siap Beraksi!")
+
+def chat_qwen(message, history):
+    # Menyusun struktur chat history
+    conversation = []
+    
+    # Masukkan history chat sebelumnya jika ada
+    for user_msg, ai_msg in history:
+        conversation.append({"role": "user", "content": user_msg})
+        conversation.append({"role": "assistant", "content": ai_msg})
+    
+    # Masukkan prompt baru dari user
+    # Tips: Tambahkan "/think" di awal prompt jika ingin mengaktifkan mode berpikir ala DeepSeek-R1
+    conversation.append({"role": "user", "content": message})
+    
+    # Convert ke format template chat Qwen
+    input_ids = tokenizer.apply_chat_template(
+        conversation, 
+        tokenize=True, 
+        add_generation_prompt=True, 
+        return_tensors="pt"
+    ).to("cpu")
+    
+    # Menggunakan Streamer biar teksnya muncul ngetik satu-satu (gak kaku)
+    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
+    
+    generation_kwargs = dict(
+        input_ids=input_ids,
+        streamer=streamer,
+        max_new_tokens=512,
+        temperature=0.6,
+        top_p=0.95
+    )
+    
+    # Jalankan generasi teks di thread terpisah agar UI Gradio gak freeze
+    thread = Thread(target=model.generate, kwargs=generation_kwargs)
+    thread.start()
+    
+    partial_text = ""
+    for new_text in streamer:
+        partial_text += new_text
+        yield partial_text
+
+# Bikin UI Chatbot yang modern pake Gradio
+demo = ui.ChatInterface(
+    fn=chat_qwen,
+    title="🤖 Qwen3-0.6B Chatbot (CPU Space)",
+    description="Berjalan lancar di HF Space Gratisan tanpa GPU. Tips: Ketik '/think' di awal pertanyaanmu untuk mengaktifkan Mode Berpikir!",
+    theme="soft"
+)
+
+if __name__ == "__main__":
+    demo.launch()