Create app.py

app.py

@@ -0,0 +1,45 @@
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+
+MODEL_ID = "janajankovic/autotrain-juhh6-uwiv9"  # your fine-tuned model repo id
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(MODEL_ID)
+
+pipe = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    max_new_tokens=256,
+    do_sample=True,
+    top_p=0.9,
+    temperature=0.7,
+)
+
+
+def chat_fn(history, user_input):
+    # simple prompt: previous turns concatenated
+    text = ""
+    for turn in history:
+        text += f"Uporabnik: {turn[0]}\nModel: {turn[1]}\n"
+    text += f"Uporabnik: {user_input}\nModel:"
+
+    out = pipe(text)[0]["generated_text"]
+    reply = out.split("Model:")[-1].strip()
+    history.append((user_input, reply))
+    return history, ""
+
+
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot(label="Slovenski finetune")
+    msg = gr.Textbox(label="Vnos")
+    clear = gr.Button("Počisti")
+
+    def user_send(message, chat_history):
+        chat_history = chat_history or []
+        return "", chat_fn(chat_history, message)[0]
+
+    msg.submit(user_send, [msg, chatbot], [msg, chatbot])
+    clear.click(lambda: ([], ""), None, [chatbot, msg], queue=False)
+
+demo.launch()