workaround

app.py

@@ -7,8 +7,9 @@ from threading import Thread
 import gradio as gr
 import spaces
 import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, pipeline
 from peft import PeftModel
+from transformers import (AutoModelForCausalLM, AutoTokenizer,
+                          TextIteratorStreamer, pipeline)
 
 DESCRIPTION = "# 真空ジェネレータ (v3)\n<p>Imitate 真空 (@vericava)'s posts interactively</p>"
 
@@ -51,11 +52,11 @@ def generate(
 
     output = my_pipeline(
         user_input,
-        temperature=temperature,
+        temperature=temperature * 1.0,
         max_new_tokens=max_new_tokens,
-        repetition_penalty=repetition_penalty,
+        repetition_penalty=repetition_penalty * 1.0,
         top_k=top_k,
-        top_p=top_p,
+        top_p=top_p * 1.0,
     )[-1]["generated_text"]
     print(output)
     gen_text = output[len(user_input):]