Update app.py

app.py

@@ -11,7 +11,8 @@ print('load ok')
 
 def completion(prompt, max_tokens, temperature, top_k, top_p):
     inpt = tokenizer.encode(prompt, return_tensors="pt")
-    out = model.generate(inpt, max_length=max_tokens, top_p=top_p, top_k=top_k, temperature=temperature)
+    tok = len(tokenizer(prompt)['input_ids'])
+    out = model.generate(inpt, max_length=tok+max_tokens, top_p=top_p, top_k=top_k, temperature=temperature)
     res = tokenizer.decode(out[0])
     return res