Create app.py

app.py

@@ -0,0 +1,36 @@
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import gradio as gr
+
+model_name = "LeoLM/leo-mistral-chat"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+)
+model = model.to("cuda" if torch.cuda.is_available() else "cpu")
+
+def chat(user_input, history=[]):
+    # Baue den Prompt auf
+    prompt = ""
+    for user, bot in history:
+        prompt += f"User: {user}\nAssistant: {bot}\n"
+    prompt += f"User: {user_input}\nAssistant:"
+
+    inputs = tokenizer(prompt, return_tensors="pt", truncation=True).to(model.device)
+    output = model.generate(**inputs, max_new_tokens=256, do_sample=True, top_p=0.9, temperature=0.7, pad_token_id=tokenizer.eos_token_id)
+    decoded = tokenizer.decode(output[0], skip_special_tokens=True)
+    answer = decoded.split("Assistant:")[-1].strip()
+    history.append((user_input, answer))
+    return history, history
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 🤖 Leichtgewichtiger KI-Chat auf Deutsch")
+    chatbot = gr.Chatbot()
+    msg = gr.Textbox(label="Deine Nachricht")
+    state = gr.State([])
+
+    msg.submit(chat, [msg, state], [chatbot, state])
+
+demo.launch()