Create app.py

Create app.py and load models: LeoLM, Occiglot, Llama2

app.py

@@ -0,0 +1,37 @@
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+# Modelle und Tokenizer laden
+model_names = {
+    "LeoLM_13B": "LeoLM/leo-hessianai-13b",
+    "Occiglot_7B": "occiglot/occiglot-7b-de-en",
+    "LLaMA2_13B": "meta-llama/Llama-2-13b-hf"
+}
+
+# Tokenizer und Modelle vorbereiten
+tokenizers = {name: AutoTokenizer.from_pretrained(model) for name, model in model_names.items()}
+models = {name: AutoModelForCausalLM.from_pretrained(model, device_map="auto", torch_dtype=torch.float16) for name, model in model_names.items()}
+
+# Funktion zur Generierung der Antwort
+def generate_response(model_choice, prompt):
+    tokenizer = tokenizers[model_choice]
+    model = models[model_choice]
+    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    outputs = model.generate(inputs["input_ids"], max_new_tokens=100, do_sample=True)
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return response
+
+# Gradio Interface
+with gr.Blocks() as demo:
+    gr.Markdown("# Vergleich von LLMs: LeoLM, Occiglot und LLaMA 2")
+    with gr.Row():
+        model_choice = gr.Radio(list(model_names.keys()), label="Modell auswählen")
+        prompt = gr.Textbox(label="Frage stellen", placeholder="Was sind die Hauptursachen für Bluthochdruck?")
+        output = gr.Textbox(label="Antwort")
+
+    submit_button = gr.Button("Antwort generieren")
+    submit_button.click(generate_response, inputs=[model_choice, prompt], outputs=output)
+
+# App starten
+demo.launch()