Update app.py

app.py

@@ -98,11 +98,12 @@ def run_interpretation(raw_interpretation_prompt, max_new_tokens, do_sample,
     
     # create an InterpretationPrompt object from raw_interpretation_prompt (after putting it in the right template)
     interpretation_prompt = global_state.interpretation_prompt_template.format(prompt=raw_interpretation_prompt, repeat=5)
-    interpretation_prompt = InterpretationPrompt(global_state.tokenizer, interpretation_prompt, layers_format=global_state.layers_format)
+    interpretation_prompt = InterpretationPrompt(global_state.tokenizer, interpretation_prompt)
 
     # generate the interpretations
     # generate = generate_interpretation_gpu if use_gpu else lambda interpretation_prompt, *args, **kwargs: interpretation_prompt.generate(*args, **kwargs)
-    generated = interpretation_prompt.generate(global_state.model, {0: interpreted_vectors}, k=3, **generation_kwargs)
+    generated = interpretation_prompt.generate(global_state.model, {0: interpreted_vectors}, layers_format=global_state.layers_format, k=3, 
+                                               **generation_kwargs)
     generation_texts = global_state.tokenizer.batch_decode(generated)
     progress_dummy_output = ''
     bubble_outputs = [gr.Textbox(text.replace('\n', ' '), visible=True, container=False, label=f'Layer {i}') for text in generation_texts]