Update app.py

app.py

@@ -1,45 +1,36 @@
-# app.py
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForCausalLM
-import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 
-# Szybszy i sensowny model
-model_name = "tiiuae/falcon-rw-1b"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+model_id = "tiiuae/falcon-rw-1b"  # Szybszy i sensowniejszy model
 
-def generate_response(user_input, history):
-    # Kontrola historii, ostatnie 3 interakcje
-    short_history = history[-3:] if history else []
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)
+generator = pipeline("text-generation", model=model, tokenizer=tokenizer)
+
+chat_history = []
+
+def chat(user_input):
+    global chat_history
+    chat_history.append(f"User: {user_input}")
+    prompt = "\n".join(chat_history) + "\nAI:"
     
-    # Budowanie promptu
-    prompt = ""
-    for human, bot in short_history:
-        prompt += f"User: {human}\nAI: {bot}\n"
-    prompt += f"User: {user_input}\nAI:"
-
-    input_ids = tokenizer.encode(prompt, return_tensors="pt", truncation=True, max_length=1024)
-    output = model.generate(
-        input_ids,
-        max_new_tokens=80,
-        do_sample=True,
-        top_k=50,
-        top_p=0.95,
-        temperature=0.7,
-        pad_token_id=tokenizer.eos_token_id
-    )
-
-    output_text = tokenizer.decode(output[0], skip_special_tokens=True)
-    reply = output_text[len(prompt):].split("User:")[0].strip()
-    history.append((user_input, reply))
-    return reply, history
-
-iface = gr.Interface(
-    fn=generate_response,
-    inputs=[gr.Textbox(label="Twoje pytanie"), gr.State([])],
-    outputs=[gr.Textbox(label="Odpowiedź AI"), gr.State([])],
-    title="🤖 Polski Chatbot AI",
-    description="Chatbot działający na bazie modelu Falcon-RW-1B. Zadaj pytanie, a AI postara się odpowiedzieć mądrze i szybko."
-)
-
-iface.launch()
+    response = generator(prompt, max_new_tokens=80, do_sample=True, top_k=40, top_p=0.9, temperature=0.7)[0]["generated_text"]
+    reply = response[len(prompt):].strip().split("\nUser:")[0]
+    
+    chat_history.append(f"AI: {reply}")
+    return reply
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 🤖 Polski Chatbot AI – Szybki i Inteligentny")
+    chatbox = gr.Chatbot()
+    msg = gr.Textbox(label="Twoja wiadomość")
+    send_btn = gr.Button("Wyślij")
+
+    def respond(user_input, history=[]):
+        reply = chat(user_input)
+        history.append((user_input, reply))
+        return history, ""
+
+    send_btn.click(respond, inputs=[msg, chatbox], outputs=[chatbox, msg])
+
+demo.launch()