Update app.py

app.py

@@ -38,7 +38,7 @@ def generate_responses(model, tokenizer, prompts):
             repetition_penalty=1.1,
             max_new_tokens=512,
         )
-    prompt_lengths = model_inputs["attention_mask"].sum(dim=1) + 1
+    prompt_lengths = model_inputs["attention_mask"].sum(dim=1) - 1
     generated_ids = [
         output_ids[length:] for length, output_ids in zip(prompt_lengths, generated_ids)
     ]